fused_with_layers.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## For Simple LSTM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from tensorflow.keras.models import Sequential\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras import layers, models\n",
    "from tensorflow.keras.layers import Dense, Flatten, Conv2D, MaxPooling2D, Dropout,InputLayer, BatchNormalization, GlobalAveragePooling2D \n",
    "from keras import models, layers\n",
    "from tensorflow.keras.layers import LSTM, Dense , Dropout\n",
    "from tensorflow.keras.losses import sparse_categorical_crossentropy\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "SEQUENCE_LENGTH = 29\n",
    "FEATURES = 51"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running for person ali\n",
      "Running for person huzaifa\n",
      "Running for person junaid\n",
      "Running for person Salman\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "\n",
    "def read_data_from_folder(folder_path, sequence_length=29):\n",
    "    class_labels = os.listdir(folder_path)\n",
    "    sequences = []\n",
    "    labels = []\n",
    "    for label_num_temp, person_folder in enumerate(class_labels):\n",
    "        print(f\"Running for person {person_folder}\")\n",
    "        person_folder_path = os.path.join(folder_path, person_folder)\n",
    "        if not os.path.isdir(person_folder_path):\n",
    "            continue\n",
    "        \n",
    "        for csv_file in os.listdir(person_folder_path):\n",
    "            if not csv_file.endswith('.csv'):\n",
    "                continue\n",
    "            \n",
    "            csv_file_path = os.path.join(person_folder_path, csv_file)\n",
    "            df = pd.read_csv(csv_file_path)\n",
    "            df = df.iloc[:, 2:]  # Select columns from the third column to the last\n",
    "            num_frames = min(sequence_length, len(df))  # Ensure not more than sequence_length frames are taken\n",
    "            \n",
    "            sequence = df.iloc[:num_frames].values\n",
    "\n",
    "            sequences.append(sequence)\n",
    "            labels.append(label_num_temp) \n",
    "\n",
    "    return sequences, labels, class_labels\n",
    "\n",
    "\n",
    "train_folder_path = 'D:/FYP/Data/video_frames/Data/CSV'\n",
    "SEQUENCE_LENGTH = 29  # Define your desired sequence length\n",
    "\n",
    "# Read data from the train folder\n",
    "sequences, labels, class_labels = read_data_from_folder(train_folder_path, sequence_length=SEQUENCE_LENGTH)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['ali', 'huzaifa', 'junaid', 'Salman']\n",
      "[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3]\n",
      "(633, 29, 51)\n",
      "The total number of classes is: 4\n"
     ]
    }
   ],
   "source": [
    "print(class_labels)\n",
    "print(labels)\n",
    "print(np.array(sequences).shape)\n",
    "# Assuming you have the labels stored in the 'labels' variable\n",
    "# Compute the number of unique labels (num_classes)\n",
    "num_classes = len(set(labels))\n",
    "print(f\"The total number of classes is: {num_classes}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{0: 6, 1: 6, 2: 9}\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "def count_occurrences(arr):\n",
    "    unique_elements, counts = np.unique(arr, return_counts=True)\n",
    "    occurrences_dict = dict(zip(unique_elements, counts))\n",
    "    return occurrences_dict\n",
    "\n",
    "# Example usage:\n",
    "labels_array = np.array([0, 0, 0, 1, 1, 1, 2, 2, 2, 2, 2, 2, 0, 0, 0, 1, 1, 1, 2, 2, 2])\n",
    "occurrences = count_occurrences(labels_array)\n",
    "print(occurrences)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{0: 147, 1: 145, 2: 138, 3: 203}\n"
     ]
    }
   ],
   "source": [
    "# Calculates the number of labels\n",
    "print(count_occurrences(labels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# print(sequences.shape)\n",
    "# print(labels.shape)\n",
    "# labels_pd = pd.DataFrame(labels)\n",
    "# print(len(labels_pd))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "sequences = np.array(sequences)\n",
    "labels = np.array(labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(474, 29, 51)\n",
      "(474,)\n",
      "(159, 29, 51)\n",
      "(159,)\n"
     ]
    }
   ],
   "source": [
    "X_Lstm, X_val_lstm, y_lstm, y_val_lstm = train_test_split(sequences, labels, test_size=0.25, random_state=42)\n",
    "print(X_Lstm.shape)\n",
    "print(y_lstm.shape)\n",
    "print(X_val_lstm.shape)\n",
    "print(y_val_lstm.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## For conv3D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import cv2\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Conv3D, MaxPooling3D, Flatten, Dense\n",
    "\n",
    "# Function to load and preprocess the video frames\n",
    "def load_video_frames(video_path, num_frames=29, image_shape=(64, 64)):\n",
    "    frames = []\n",
    "    cap = cv2.VideoCapture(video_path)\n",
    "\n",
    "    # Get total frames and calculate skip_frames to sample equally from the video\n",
    "    total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))\n",
    "    # skip_frames = max(total_frames // num_frames, 1)\n",
    "\n",
    "    # Loop through the frames with the adjusted skip_frames\n",
    "    for i in range(0, total_frames):\n",
    "        cap.set(cv2.CAP_PROP_POS_FRAMES, i)\n",
    "        ret, frame = cap.read()\n",
    "        if ret:\n",
    "            frame = cv2.resize(frame, image_shape)\n",
    "            frames.append(frame)\n",
    "\n",
    "    cap.release()\n",
    "\n",
    "    # Pad frames if the video has fewer frames than required\n",
    "    if len(frames) < num_frames:\n",
    "        print(\"Length of frames is \",len(frames))\n",
    "        print(\"Number of frames is \",num_frames)\n",
    "        frames.extend([frames[-1]] * (num_frames - len(frames)))\n",
    "        # indices = np.linspace(0, len(frames) - 1, num_frames, dtype=np.int)\n",
    "        # frames = [frames[i] for i in indices]    \n",
    "    elif len(frames) > num_frames:\n",
    "        # If the number of frames is greater, perform downsampling\n",
    "        downsample_factor = len(frames) // num_frames\n",
    "        indices = np.arange(0, len(frames), downsample_factor)[:num_frames]\n",
    "        frames = [frames[i] for i in indices]\n",
    "    print(\"Length of frames is \",len(frames))\n",
    "    return np.array(frames)\n",
    "\n",
    "# Function to load all videos and their labels from the root folder\n",
    "def load_data(root_folder):\n",
    "    class_labels = os.listdir(root_folder)\n",
    "    videos = []\n",
    "    labels = []\n",
    "\n",
    "    for i, class_label in enumerate(class_labels):\n",
    "        class_folder = os.path.join(root_folder, class_label)\n",
    "        video_files = os.listdir(class_folder)\n",
    "\n",
    "        for video_file in video_files:\n",
    "            video_path = os.path.join(class_folder, video_file)\n",
    "            frames = load_video_frames(video_path)\n",
    "            videos.append(frames)\n",
    "            print( np.array(videos).shape )\n",
    "\n",
    "            labels.append(i)\n",
    "\n",
    "    return np.array(videos), np.array(labels) , np.array(class_labels)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Length of frames is  29\n",
      "(1, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(2, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(3, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(4, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(5, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(6, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(7, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(8, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(9, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(10, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(11, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(12, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(13, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(14, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(15, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(16, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(17, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(18, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(19, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(20, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(21, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(22, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(23, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(24, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(25, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(26, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(27, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(28, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(29, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(30, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(31, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(32, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(33, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(34, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(35, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(36, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(37, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(38, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(39, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(40, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(41, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(42, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(43, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(44, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(45, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(46, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(47, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(48, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(49, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(50, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(51, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(52, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(53, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(54, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(55, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(56, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(57, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(58, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(59, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(60, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(61, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(62, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(63, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(64, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(65, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(66, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(67, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(68, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(69, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(70, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(71, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(72, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(73, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(74, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(75, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(76, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(77, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(78, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(79, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(80, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(81, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(82, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(83, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(84, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(85, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(86, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(87, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(88, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(89, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(90, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(91, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(92, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(93, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(94, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(95, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(96, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(97, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(98, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(99, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(100, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(101, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(102, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(103, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(104, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(105, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(106, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(107, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(108, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(109, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(110, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(111, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(112, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(113, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(114, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(115, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(116, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(117, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(118, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(119, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(120, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(121, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(122, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(123, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(124, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(125, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(126, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(127, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(128, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(129, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(130, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(131, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(132, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(133, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(134, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(135, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(136, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(137, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(138, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(139, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(140, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(141, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(142, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(143, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(144, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(145, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(146, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(147, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(148, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(149, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(150, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(151, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(152, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(153, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(154, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(155, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(156, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(157, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(158, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(159, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(160, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(161, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(162, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(163, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(164, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(165, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(166, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(167, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(168, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(169, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(170, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(171, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(172, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(173, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(174, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(175, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(176, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(177, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(178, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(179, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(180, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(181, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(182, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(183, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(184, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(185, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(186, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(187, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(188, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(189, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(190, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(191, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(192, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(193, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(194, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(195, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(196, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(197, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(198, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(199, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(200, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(201, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(202, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(203, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(204, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(205, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(206, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(207, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(208, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(209, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(210, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(211, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(212, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(213, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(214, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(215, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(216, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(217, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(218, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(219, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(220, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(221, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(222, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(223, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(224, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(225, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(226, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(227, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(228, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(229, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(230, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(231, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(232, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(233, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(234, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(235, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(236, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(237, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(238, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(239, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(240, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(241, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(242, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(243, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(244, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(245, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(246, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(247, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(248, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(249, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(250, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(251, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(252, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(253, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(254, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(255, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(256, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(257, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(258, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(259, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(260, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(261, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(262, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(263, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(264, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(265, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(266, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(267, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(268, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(269, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(270, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(271, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(272, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(273, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(274, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(275, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(276, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(277, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(278, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(279, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(280, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(281, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(282, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(283, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(284, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(285, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(286, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(287, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(288, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(289, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(290, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(291, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(292, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(293, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(294, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(295, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(296, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(297, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(298, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(299, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(300, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(301, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(302, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(303, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(304, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(305, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(306, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(307, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(308, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(309, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(310, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(311, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(312, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(313, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(314, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(315, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(316, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(317, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(318, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(319, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(320, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(321, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(322, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(323, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(324, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(325, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(326, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(327, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(328, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(329, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(330, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(331, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(332, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(333, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(334, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(335, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(336, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(337, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(338, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(339, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(340, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(341, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(342, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(343, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(344, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(345, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(346, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(347, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(348, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(349, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(350, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(351, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(352, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(353, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(354, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(355, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(356, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(357, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(358, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(359, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(360, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(361, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(362, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(363, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(364, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(365, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(366, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(367, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(368, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(369, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(370, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(371, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(372, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(373, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(374, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(375, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(376, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(377, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(378, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(379, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(380, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(381, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(382, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(383, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(384, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(385, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(386, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(387, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(388, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(389, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(390, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(391, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(392, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(393, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(394, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(395, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(396, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(397, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(398, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(399, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(400, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(401, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(402, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(403, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(404, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(405, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(406, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(407, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(408, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(409, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(410, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(411, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(412, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(413, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(414, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(415, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(416, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(417, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(418, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(419, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(420, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(421, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(422, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(423, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(424, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(425, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(426, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(427, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(428, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(429, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(430, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(431, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(432, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(433, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(434, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(435, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(436, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(437, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(438, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(439, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(440, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(441, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(442, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(443, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(444, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(445, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(446, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(447, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(448, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(449, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(450, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(451, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(452, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(453, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(454, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(455, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(456, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(457, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(458, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(459, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(460, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(461, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(462, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(463, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(464, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(465, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(466, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(467, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(468, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(469, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(470, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(471, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(472, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(473, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(474, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(475, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(476, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(477, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(478, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(479, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(480, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(481, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(482, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(483, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(484, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(485, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(486, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(487, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(488, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(489, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(490, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(491, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(492, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(493, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(494, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(495, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(496, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(497, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(498, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(499, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(500, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(501, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(502, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(503, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(504, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(505, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(506, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(507, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(508, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(509, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(510, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(511, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(512, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(513, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(514, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(515, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(516, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(517, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(518, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(519, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(520, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(521, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(522, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(523, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(524, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(525, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(526, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(527, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(528, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(529, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(530, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(531, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(532, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(533, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(534, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(535, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(536, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(537, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(538, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(539, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(540, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(541, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(542, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(543, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(544, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(545, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(546, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(547, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(548, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(549, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(550, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(551, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(552, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(553, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(554, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(555, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(556, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(557, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(558, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(559, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(560, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(561, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(562, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(563, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(564, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(565, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(566, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(567, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(568, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(569, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(570, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(571, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(572, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(573, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(574, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(575, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(576, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(577, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(578, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(579, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(580, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(581, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(582, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(583, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(584, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(585, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(586, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(587, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(588, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(589, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(590, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(591, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(592, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(593, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(594, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(595, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(596, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(597, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(598, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(599, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(600, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(601, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(602, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(603, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(604, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(605, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(606, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(607, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(608, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(609, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(610, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(611, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(612, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(613, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(614, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(615, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(616, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(617, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(618, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(619, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(620, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(621, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(622, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(623, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(624, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(625, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(626, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(627, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(628, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(629, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(630, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(631, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(632, 29, 64, 64, 3)\n",
      "Length of frames is  29\n",
      "(633, 29, 64, 64, 3)\n"
     ]
    }
   ],
   "source": [
    "root_folder = 'D:/FYP/Data/video_frames/Data/1sec_videos'\n",
    "with tf.device(\"CPU\"):\n",
    "    videos, labels , class_label = load_data(root_folder)\n",
    "    # Split data into training and testing sets\n",
    "    X_conv3d, X_val_conv3d, y_conv3d, y_val_conv3d = train_test_split(videos, labels, test_size=0.25, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Class Labels:  ['ali' 'huzaifa' 'junaid' 'salman']\n",
      "<class 'numpy.ndarray'>\n",
      "<class 'numpy.ndarray'>\n",
      "<class 'numpy.ndarray'>\n",
      "<class 'numpy.ndarray'>\n",
      "The shape of X_conv3d is:  (474, 29, 64, 64, 3)\n",
      "The shape of y_conv3d is:  (474,)\n",
      "The total number of labels is:  4\n"
     ]
    }
   ],
   "source": [
    "# y_conv3d = y_conv3d.reshape(-1, 1)\n",
    "# y_val_conv3d = y_val_conv3d.reshape(-1, 1)\n",
    "print(\"Class Labels: \", class_label)\n",
    "print(type(X_conv3d))\n",
    "print(type(y_conv3d))\n",
    "print(type(X_val_conv3d))\n",
    "print(type(y_val_conv3d))\n",
    "print(\"The shape of X_conv3d is: \", X_conv3d.shape)\n",
    "print(\"The shape of y_conv3d is: \", y_conv3d.shape)\n",
    "print(\"The total number of labels is: \", len(np.unique(labels)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(29, 64, 64, 3)\n"
     ]
    }
   ],
   "source": [
    "input_shape = X_conv3d.shape[1:]\n",
    "print(input_shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.layers import Input, Dense, LSTM, Conv3D, Flatten\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import make_pipeline\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fusion "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# ---------------------- Fusion Without Layers ---------------------#\n",
    "# def simple_model(input_shape, num_classes):\n",
    "#     # Create the Conv3D model\n",
    "#     model = tf.keras.Sequential([\n",
    "#         Conv3D(2, kernel_size=(3, 3, 3), activation='relu', input_shape=input_shape),\n",
    "#         Flatten(),  # Flatten the 3D convolution output\n",
    "#         Dense(num_classes, activation='softmax')\n",
    "#     ])\n",
    "#     return model\n",
    "\n",
    "# def simple_lstm(sequence_length, features, num_classes):\n",
    "#     model = tf.keras.Sequential()\n",
    "#     model.add(LSTM(64, input_shape=(sequence_length, features)))  \n",
    "#     model.add(Dense(num_classes, activation='softmax'))  # num_classes = number of unique labels\n",
    "#     return model\n",
    "\n",
    "# def late_fusion_model(conv3d_input_shape, lstm_input_shape, num_classes):\n",
    "#     conv3d_input = Input(shape=conv3d_input_shape)\n",
    "#     lstm_input = Input(shape=lstm_input_shape)\n",
    "\n",
    "#     conv3d_model = simple_model(conv3d_input_shape, num_classes)\n",
    "#     lstm_model = simple_lstm(lstm_input_shape[0], lstm_input_shape[1], num_classes)\n",
    "#     # Load pre-trained weights for the Conv3D model\n",
    "#     print(\"Loading Conv3D model.......\")\n",
    "#     conv3d_model.load_weights(\"../simple_model_weights.h5\")\n",
    "\n",
    "#     print(\"Loading LSTM model.........\")\n",
    "#     # Load pre-trained weights for the LSTM model\n",
    "#     lstm_model.load_weights(\"./lstm_model.h5\")\n",
    "\n",
    "#     conv3d_output = conv3d_model(conv3d_input)\n",
    "#     lstm_output = lstm_model(lstm_input)\n",
    "\n",
    "#     print(\"Fusing the models......\")\n",
    "#     # Perform late fusion by concatenating the probabilities from both models\n",
    "#     fused_output = tf.keras.layers.Concatenate()([conv3d_output, lstm_output])\n",
    "\n",
    "#     fused_model = tf.keras.Model(inputs=[conv3d_input, lstm_input], outputs=fused_output)\n",
    "#     fused_model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "#     fused_model.summary()\n",
    "#     return fused_model\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "# ---------------------- Fusion With Layers ---------------------#\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.layers import Conv3D, Flatten, Dense, LSTM, Input, Concatenate\n",
    "\n",
    "def simple_model(input_shape, num_classes):\n",
    "    model = tf.keras.Sequential([\n",
    "        Conv3D(2, kernel_size=(3, 3, 3), activation='relu', input_shape=input_shape),\n",
    "        Flatten(),\n",
    "        Dense(num_classes, activation='softmax')\n",
    "    ])\n",
    "    return model\n",
    "\n",
    "def simple_lstm(sequence_length, features, num_classes):\n",
    "    model = tf.keras.Sequential()\n",
    "    model.add(LSTM(64, input_shape=(sequence_length, features)))\n",
    "    model.add(Dense(num_classes, activation='softmax'))\n",
    "    return model\n",
    "\n",
    "def late_fusion_model(conv3d_input_shape, lstm_input_shape, num_classes):\n",
    "    conv3d_input = Input(shape=conv3d_input_shape)\n",
    "    lstm_input = Input(shape=lstm_input_shape)\n",
    "\n",
    "    conv3d_model = simple_model(conv3d_input_shape, num_classes)\n",
    "    lstm_model = simple_lstm(lstm_input_shape[0], lstm_input_shape[1], num_classes)\n",
    "\n",
    "    # Load pre-trained weights for the Conv3D model\n",
    "    # conv3d_model.load_weights(\"../simple_model_weights.h5\")\n",
    "\n",
    "    # Load pre-trained weights for the LSTM model\n",
    "    # lstm_model.load_weights(\"./lstm_model.h5\")\n",
    "\n",
    "    conv3d_output = conv3d_model(conv3d_input)\n",
    "    lstm_output = lstm_model(lstm_input)\n",
    "\n",
    "    # Perform late fusion by concatenating the probabilities from both models\n",
    "    fused_output = Concatenate()([conv3d_output, lstm_output])\n",
    "    fused_output = Dense(num_classes, activation='softmax')(fused_output)  # Additional layer for fusion\n",
    "\n",
    "    fused_model = tf.keras.Model(inputs=[conv3d_input, lstm_input], outputs=fused_output)\n",
    "    fused_model.compile(loss='sparse_categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "    fused_model.summary()\n",
    "    return fused_model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The shape of X_conv3D:  (474, 29, 64, 64, 3)\n",
      "The shape of y_conv3D:  (474,)\n",
      "The shape of X_Lstm:  (474, 29, 51)\n",
      "The shape of y_Lstm:  (474,)\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Example usage\n",
    "conv3d_input_shape = (29, 64, 64, 3)\n",
    "lstm_input_shape = (29, 51)\n",
    "num_classes = 4\n",
    "print(\"The shape of X_conv3D: \",X_conv3d.shape)\n",
    "print(\"The shape of y_conv3D: \",y_conv3d.shape)\n",
    "print(\"The shape of X_Lstm: \",X_Lstm.shape)\n",
    "print(\"The shape of y_Lstm: \",y_lstm.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[3 2 1 2 2 0 3 1 1 0 3 3 2 3 1 0 1 2 3 1 1 3 3 1 1 1 0 0 3 0]\n",
      "[3 2 1 2 2 0 3 1 1 0 3 3 2 3 1 0 1 2 3 1 1 3 3 1 1 1 0 0 3 0]\n"
     ]
    }
   ],
   "source": [
    "print(y_conv3d[0:30])\n",
    "print(y_lstm[0:30])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Assuming you have training data X_conv3d and X_Lstm, and labels y\n",
    "# num_samples_conv3d = X_conv3d.shape[0]\n",
    "# print(num_samples_conv3d)\n",
    "# subsample_indices = np.random.choice(X_Lstm.shape[0], num_samples_conv3d, replace=False)\n",
    "# print(subsample_indices)\n",
    "# X_Lstm_subsampled = X_Lstm[subsample_indices]\n",
    "# y_lstm_subsampled = y_lstm[subsample_indices]\n",
    "# print(\"The shape of X_Lstm_subsampled: \",X_Lstm_subsampled.shape)\n",
    "# print(\"The shape of Y_Lstm_subsampled: \",y_lstm_subsampled.shape)\n",
    "# print(\"The shape of X_conv3D: \",X_conv3d.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# print(y_conv3d)\n",
    "# print(y_lstm_subsampled)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(29, 64, 64, 3)\n",
      "(29, 51)\n",
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                   Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      " input_1 (InputLayer)           [(None, 29, 64, 64,  0           []                               \n",
      "                                 3)]                                                              \n",
      "                                                                                                  \n",
      " input_2 (InputLayer)           [(None, 29, 51)]     0           []                               \n",
      "                                                                                                  \n",
      " sequential (Sequential)        (None, 4)            830472      ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_1 (Sequential)      (None, 4)            29956       ['input_2[0][0]']                \n",
      "                                                                                                  \n",
      " concatenate (Concatenate)      (None, 8)            0           ['sequential[0][0]',             \n",
      "                                                                  'sequential_1[0][0]']           \n",
      "                                                                                                  \n",
      " dense_2 (Dense)                (None, 4)            36          ['concatenate[0][0]']            \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 860,464\n",
      "Trainable params: 860,464\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "print(conv3d_input_shape)\n",
    "print(lstm_input_shape)\n",
    "fused_model = late_fusion_model(conv3d_input_shape, lstm_input_shape, num_classes)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_val_conv3d shape: (159, 29, 64, 64, 3)\n",
      "X_val_lstm shape: (159, 29, 51)\n",
      "y_val_conv3d shape: (159,)\n",
      "X_val_conv3d dtype: uint8\n",
      "X_val_lstm dtype: float32\n",
      "y_val_conv3d dtype: int32\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "# Convert X_val_lstm to float32\n",
    "X_val_lstm = X_val_lstm.astype(np.float32)\n",
    "\n",
    "# Convert y_val_conv3d to int32\n",
    "y_val_conv3d = y_val_conv3d.astype(np.int32)\n",
    "\n",
    "print(\"X_val_conv3d shape:\", X_val_conv3d.shape)\n",
    "print(\"X_val_lstm shape:\", X_val_lstm.shape)\n",
    "print(\"y_val_conv3d shape:\", y_val_conv3d.shape)\n",
    "print(\"X_val_conv3d dtype:\", X_val_conv3d.dtype)\n",
    "print(\"X_val_lstm dtype:\", X_val_lstm.dtype)\n",
    "print(\"y_val_conv3d dtype:\", y_val_conv3d.dtype)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "15/15 [==============================] - 22s 500ms/step - loss: 1.4888 - accuracy: 0.2996 - val_loss: 1.3883 - val_accuracy: 0.3836\n",
      "Epoch 2/100\n",
      "15/15 [==============================] - 2s 150ms/step - loss: 1.4533 - accuracy: 0.2996 - val_loss: 1.3758 - val_accuracy: 0.3836\n",
      "Epoch 3/100\n",
      "15/15 [==============================] - 2s 110ms/step - loss: 1.4315 - accuracy: 0.2996 - val_loss: 1.3663 - val_accuracy: 0.3836\n",
      "Epoch 4/100\n",
      "15/15 [==============================] - 1s 91ms/step - loss: 1.4151 - accuracy: 0.2996 - val_loss: 1.3555 - val_accuracy: 0.3836\n",
      "Epoch 5/100\n",
      "15/15 [==============================] - 1s 83ms/step - loss: 1.4003 - accuracy: 0.2996 - val_loss: 1.3459 - val_accuracy: 0.3836\n",
      "Epoch 6/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 1.3860 - accuracy: 0.2996 - val_loss: 1.3320 - val_accuracy: 0.3836\n",
      "Epoch 7/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 1.3678 - accuracy: 0.2996 - val_loss: 1.3155 - val_accuracy: 0.3836\n",
      "Epoch 8/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 1.3572 - accuracy: 0.2996 - val_loss: 1.2991 - val_accuracy: 0.3836\n",
      "Epoch 9/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.3356 - accuracy: 0.2996 - val_loss: 1.2883 - val_accuracy: 0.3836\n",
      "Epoch 10/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.3316 - accuracy: 0.2996 - val_loss: 1.2821 - val_accuracy: 0.3836\n",
      "Epoch 11/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 1.3093 - accuracy: 0.2996 - val_loss: 1.2617 - val_accuracy: 0.3836\n",
      "Epoch 12/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.2921 - accuracy: 0.2996 - val_loss: 1.2468 - val_accuracy: 0.3836\n",
      "Epoch 13/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.2758 - accuracy: 0.2996 - val_loss: 1.2289 - val_accuracy: 0.3836\n",
      "Epoch 14/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 1.2633 - accuracy: 0.3502 - val_loss: 1.2413 - val_accuracy: 0.4465\n",
      "Epoch 15/100\n",
      "15/15 [==============================] - 1s 83ms/step - loss: 1.2648 - accuracy: 0.3629 - val_loss: 1.2276 - val_accuracy: 0.4277\n",
      "Epoch 16/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.2463 - accuracy: 0.5654 - val_loss: 1.2007 - val_accuracy: 0.6226\n",
      "Epoch 17/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 1.2269 - accuracy: 0.6097 - val_loss: 1.1874 - val_accuracy: 0.6289\n",
      "Epoch 18/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 1.2062 - accuracy: 0.6329 - val_loss: 1.1728 - val_accuracy: 0.6478\n",
      "Epoch 19/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.2010 - accuracy: 0.6329 - val_loss: 1.1603 - val_accuracy: 0.6415\n",
      "Epoch 20/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.1807 - accuracy: 0.6392 - val_loss: 1.1731 - val_accuracy: 0.6478\n",
      "Epoch 21/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.1768 - accuracy: 0.6392 - val_loss: 1.1515 - val_accuracy: 0.6478\n",
      "Epoch 22/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 1.1612 - accuracy: 0.6392 - val_loss: 1.1312 - val_accuracy: 0.6478\n",
      "Epoch 23/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 1.1456 - accuracy: 0.6392 - val_loss: 1.1256 - val_accuracy: 0.6478\n",
      "Epoch 24/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 1.1283 - accuracy: 0.6392 - val_loss: 1.0986 - val_accuracy: 0.6478\n",
      "Epoch 25/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 1.1163 - accuracy: 0.6392 - val_loss: 1.0866 - val_accuracy: 0.6478\n",
      "Epoch 26/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 1.1024 - accuracy: 0.6392 - val_loss: 1.0644 - val_accuracy: 0.6478\n",
      "Epoch 27/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.0734 - accuracy: 0.6392 - val_loss: 1.0522 - val_accuracy: 0.6478\n",
      "Epoch 28/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 1.0579 - accuracy: 0.6392 - val_loss: 1.0365 - val_accuracy: 0.6478\n",
      "Epoch 29/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 1.0439 - accuracy: 0.6392 - val_loss: 1.0173 - val_accuracy: 0.6478\n",
      "Epoch 30/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 1.0279 - accuracy: 0.6392 - val_loss: 0.9935 - val_accuracy: 0.6478\n",
      "Epoch 31/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 1.0348 - accuracy: 0.6392 - val_loss: 1.0232 - val_accuracy: 0.6478\n",
      "Epoch 32/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 1.0194 - accuracy: 0.6392 - val_loss: 0.9909 - val_accuracy: 0.6478\n",
      "Epoch 33/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.9963 - accuracy: 0.6392 - val_loss: 0.9776 - val_accuracy: 0.6478\n",
      "Epoch 34/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.9846 - accuracy: 0.6392 - val_loss: 0.9765 - val_accuracy: 0.6478\n",
      "Epoch 35/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.9702 - accuracy: 0.6392 - val_loss: 0.9450 - val_accuracy: 0.6478\n",
      "Epoch 36/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.9507 - accuracy: 0.6392 - val_loss: 0.9314 - val_accuracy: 0.6478\n",
      "Epoch 37/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 0.9358 - accuracy: 0.6392 - val_loss: 0.9186 - val_accuracy: 0.6478\n",
      "Epoch 38/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.9208 - accuracy: 0.6414 - val_loss: 0.8994 - val_accuracy: 0.6478\n",
      "Epoch 39/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.9044 - accuracy: 0.6857 - val_loss: 0.8848 - val_accuracy: 0.6478\n",
      "Epoch 40/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.8851 - accuracy: 0.6519 - val_loss: 0.8667 - val_accuracy: 0.6478\n",
      "Epoch 41/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.8681 - accuracy: 0.7342 - val_loss: 0.8517 - val_accuracy: 0.8176\n",
      "Epoch 42/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.8535 - accuracy: 0.8143 - val_loss: 0.8397 - val_accuracy: 0.8239\n",
      "Epoch 43/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.8380 - accuracy: 0.7869 - val_loss: 0.8206 - val_accuracy: 0.7925\n",
      "Epoch 44/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.8220 - accuracy: 0.7954 - val_loss: 0.8088 - val_accuracy: 0.8176\n",
      "Epoch 45/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.8056 - accuracy: 0.8207 - val_loss: 0.7911 - val_accuracy: 0.8113\n",
      "Epoch 46/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.7902 - accuracy: 0.8481 - val_loss: 0.7758 - val_accuracy: 0.8553\n",
      "Epoch 47/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.7755 - accuracy: 0.8586 - val_loss: 0.7645 - val_accuracy: 0.8553\n",
      "Epoch 48/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.7624 - accuracy: 0.8586 - val_loss: 0.7504 - val_accuracy: 0.8553\n",
      "Epoch 49/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.7493 - accuracy: 0.8608 - val_loss: 0.7367 - val_accuracy: 0.8553\n",
      "Epoch 50/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.7355 - accuracy: 0.8586 - val_loss: 0.7272 - val_accuracy: 0.8553\n",
      "Epoch 51/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.7243 - accuracy: 0.8608 - val_loss: 0.7142 - val_accuracy: 0.8553\n",
      "Epoch 52/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.7108 - accuracy: 0.8608 - val_loss: 0.6985 - val_accuracy: 0.8553\n",
      "Epoch 53/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.6991 - accuracy: 0.8608 - val_loss: 0.6883 - val_accuracy: 0.8553\n",
      "Epoch 54/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.6858 - accuracy: 0.8586 - val_loss: 0.6744 - val_accuracy: 0.8553\n",
      "Epoch 55/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.6733 - accuracy: 0.8608 - val_loss: 0.6631 - val_accuracy: 0.8553\n",
      "Epoch 56/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.6631 - accuracy: 0.8608 - val_loss: 0.6537 - val_accuracy: 0.8553\n",
      "Epoch 57/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.6521 - accuracy: 0.8608 - val_loss: 0.6422 - val_accuracy: 0.8553\n",
      "Epoch 58/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.6410 - accuracy: 0.8608 - val_loss: 0.6325 - val_accuracy: 0.8553\n",
      "Epoch 59/100\n",
      "15/15 [==============================] - 1s 78ms/step - loss: 0.6311 - accuracy: 0.8608 - val_loss: 0.6226 - val_accuracy: 0.8553\n",
      "Epoch 60/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 0.6226 - accuracy: 0.8608 - val_loss: 0.6145 - val_accuracy: 0.8553\n",
      "Epoch 61/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.6131 - accuracy: 0.8608 - val_loss: 0.6041 - val_accuracy: 0.8553\n",
      "Epoch 62/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.6031 - accuracy: 0.8608 - val_loss: 0.5954 - val_accuracy: 0.8553\n",
      "Epoch 63/100\n",
      "15/15 [==============================] - 1s 83ms/step - loss: 0.5943 - accuracy: 0.8608 - val_loss: 0.5869 - val_accuracy: 0.8553\n",
      "Epoch 64/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.5860 - accuracy: 0.8608 - val_loss: 0.5808 - val_accuracy: 0.8553\n",
      "Epoch 65/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.5790 - accuracy: 0.8608 - val_loss: 0.5721 - val_accuracy: 0.8553\n",
      "Epoch 66/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.5712 - accuracy: 0.8608 - val_loss: 0.5634 - val_accuracy: 0.8553\n",
      "Epoch 67/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.5626 - accuracy: 0.8608 - val_loss: 0.5563 - val_accuracy: 0.8553\n",
      "Epoch 68/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.5554 - accuracy: 0.8608 - val_loss: 0.5486 - val_accuracy: 0.8553\n",
      "Epoch 69/100\n",
      "15/15 [==============================] - 1s 83ms/step - loss: 0.5483 - accuracy: 0.8608 - val_loss: 0.5425 - val_accuracy: 0.8553\n",
      "Epoch 70/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 0.5416 - accuracy: 0.8608 - val_loss: 0.5362 - val_accuracy: 0.8553\n",
      "Epoch 71/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 0.5353 - accuracy: 0.8608 - val_loss: 0.5294 - val_accuracy: 0.8553\n",
      "Epoch 72/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.5281 - accuracy: 0.8608 - val_loss: 0.5231 - val_accuracy: 0.8553\n",
      "Epoch 73/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.5218 - accuracy: 0.8608 - val_loss: 0.5167 - val_accuracy: 0.8553\n",
      "Epoch 74/100\n",
      "15/15 [==============================] - 1s 83ms/step - loss: 0.5157 - accuracy: 0.8608 - val_loss: 0.5118 - val_accuracy: 0.8553\n",
      "Epoch 75/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.5106 - accuracy: 0.8608 - val_loss: 0.5053 - val_accuracy: 0.8553\n",
      "Epoch 76/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.5043 - accuracy: 0.8608 - val_loss: 0.5009 - val_accuracy: 0.8553\n",
      "Epoch 77/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4997 - accuracy: 0.8608 - val_loss: 0.4945 - val_accuracy: 0.8553\n",
      "Epoch 78/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4934 - accuracy: 0.8608 - val_loss: 0.4894 - val_accuracy: 0.8553\n",
      "Epoch 79/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4887 - accuracy: 0.8608 - val_loss: 0.4842 - val_accuracy: 0.8553\n",
      "Epoch 80/100\n",
      "15/15 [==============================] - 1s 85ms/step - loss: 0.4832 - accuracy: 0.8608 - val_loss: 0.4790 - val_accuracy: 0.8553\n",
      "Epoch 81/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4795 - accuracy: 0.8608 - val_loss: 0.4762 - val_accuracy: 0.8553\n",
      "Epoch 82/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4752 - accuracy: 0.8608 - val_loss: 0.4707 - val_accuracy: 0.8553\n",
      "Epoch 83/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4704 - accuracy: 0.8608 - val_loss: 0.4661 - val_accuracy: 0.8553\n",
      "Epoch 84/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4658 - accuracy: 0.8608 - val_loss: 0.4631 - val_accuracy: 0.8553\n",
      "Epoch 85/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4635 - accuracy: 0.8608 - val_loss: 0.4604 - val_accuracy: 0.8553\n",
      "Epoch 86/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4587 - accuracy: 0.8608 - val_loss: 0.4584 - val_accuracy: 0.8553\n",
      "Epoch 87/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.4541 - accuracy: 0.8608 - val_loss: 0.4523 - val_accuracy: 0.8553\n",
      "Epoch 88/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.4509 - accuracy: 0.8608 - val_loss: 0.4454 - val_accuracy: 0.8553\n",
      "Epoch 89/100\n",
      "15/15 [==============================] - 1s 82ms/step - loss: 0.4445 - accuracy: 0.8608 - val_loss: 0.4407 - val_accuracy: 0.8553\n",
      "Epoch 90/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.4407 - accuracy: 0.8608 - val_loss: 0.4371 - val_accuracy: 0.8553\n",
      "Epoch 91/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.4371 - accuracy: 0.8608 - val_loss: 0.4345 - val_accuracy: 0.8553\n",
      "Epoch 92/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4340 - accuracy: 0.8608 - val_loss: 0.4307 - val_accuracy: 0.8553\n",
      "Epoch 93/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4303 - accuracy: 0.8608 - val_loss: 0.4268 - val_accuracy: 0.8553\n",
      "Epoch 94/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.4268 - accuracy: 0.8608 - val_loss: 0.4233 - val_accuracy: 0.8553\n",
      "Epoch 95/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4232 - accuracy: 0.8608 - val_loss: 0.4204 - val_accuracy: 0.8553\n",
      "Epoch 96/100\n",
      "15/15 [==============================] - 1s 79ms/step - loss: 0.4201 - accuracy: 0.8608 - val_loss: 0.4171 - val_accuracy: 0.8553\n",
      "Epoch 97/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4171 - accuracy: 0.8608 - val_loss: 0.4145 - val_accuracy: 0.8553\n",
      "Epoch 98/100\n",
      "15/15 [==============================] - 1s 81ms/step - loss: 0.4144 - accuracy: 0.8608 - val_loss: 0.4109 - val_accuracy: 0.8553\n",
      "Epoch 99/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4110 - accuracy: 0.8608 - val_loss: 0.4077 - val_accuracy: 0.8553\n",
      "Epoch 100/100\n",
      "15/15 [==============================] - 1s 80ms/step - loss: 0.4083 - accuracy: 0.8608 - val_loss: 0.4057 - val_accuracy: 0.8553\n"
     ]
    }
   ],
   "source": [
    "firtting = fused_model.fit([X_conv3d, X_Lstm], y_lstm,\n",
    "                          validation_data=([X_val_conv3d, X_val_lstm], y_val_conv3d),\n",
    "                          epochs=100, verbose=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "fused_model.save_weights('fused_model_conv(29,64,64,3)_lstm(29,51).h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAzoAAANBCAYAAAAoVMu2AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAAChDklEQVR4nOzddZiV1d6H8Xvv6WCG7u5uFcUWRRRUbEzs7uTY3Z3HbmxERVRQQQQVkZbubplhYJh+/9jKkZeQgdmzJ+7Pde2LNXut51m/4Zz3vH5d61lPoKCgoABJkiRJKkOCkS5AkiRJkoqaQUeSJElSmWPQkSRJklTmGHQkSZIklTkGHUmSJElljkFHkiRJUplj0JEkSZJU5hh0JEmSJJU50ZEuYFfk5+ezbNkyKlSoQCAQiHQ5kiRJkiKkoKCADRs2ULt2bYLBHa/blIqgs2zZMurVqxfpMiRJkiSVEIsXL6Zu3bo77C8VQadChQpA6JdJSUmJcDWSJEmSIiU9PZ169eptyQg7UiqCzt/b1VJSUgw6kiRJkv71kRYPI5AkSZJU5hh0JEmSJJU5Bh1JkiRJZU6peEZnV+Tl5ZGTkxPpMrSHoqKiiI6O9hhxSZIk7ZEyEXQyMjJYsmQJBQUFkS5FRSAxMZFatWoRGxsb6VIkSZJUSpX6oJOXl8eSJUtITEykWrVqrgSUYgUFBWRnZ7N69Wrmz59Ps2bNdvoSKEmSJGlHSn3QycnJoaCggGrVqpGQkBDpcrSHEhISiImJYeHChWRnZxMfHx/pkiRJklQKlZl/Xe5KTtnhKo4kSZL2lP9EKUmSJKnMMehIkiRJKnMMOmVEw4YNefLJJyNdhiRJklQiGHQi5OCDD+bqq68usvv99ttvXHjhhXt0j6KuSZIkSYqUUn/qWllWUFBAXl4e0dH//h9TtWrViqEiSZIkqXQocys6BQUFbMrOjchnV19Y2r9/f0aOHMlTTz1FIBAgEAiwYMECRowYQSAQYOjQoXTp0oW4uDh++ukn5s6dy7HHHkuNGjVITk5mr732Yvjw4Vvd8/9vXQsEArzyyiv07duXxMREmjVrxueff75Hf7effPIJbdq0IS4ujoYNG/LYY49t1f/888/TrFkz4uPjqVGjBieeeOKWvo8//ph27dqRkJBAlSpV6NGjBxs3btyjeiRJkqQdKXMrOpk5ebS+/ZuIzD3t7p4kxv77X+lTTz3FrFmzaNu2LXfffTcQWpFZsGABADfffDOPPvoojRs3plKlSixevJijjjqK++67j7i4ON566y369OnDzJkzqV+//g7nueuuu3j44Yd55JFHeOaZZzj99NNZuHAhlStXLvTv9vvvv3PyySdz5513csoppzBmzBguvfRSqlSpQv/+/Rk3bhxXXnklb7/9Nvvttx/r1q1j1KhRACxfvpx+/frx8MMP07dvXzZs2MCoUaN2ORhKkiRJhVXmgk5pkJqaSmxsLImJidSsWXOb/rvvvpvDDz98y8+VK1emQ4cOW36+5557GDRoEJ9//jmXX375Dufp378//fr1A+D+++/n6aefZuzYsRx55JGFrvnxxx/nsMMO47bbbgOgefPmTJs2jUceeYT+/fuzaNEikpKS6N27NxUqVKBBgwZ06tQJCAWd3Nxcjj/+eBo0aABAu3btCl2DJEmStKvKXNBJiIli2t09IzZ3UejatetWP2dkZHDnnXcyZMiQLaEhMzOTRYsW7fQ+7du339JOSkoiJSWFVatW7VZN06dP59hjj93qu+7du/Pkk0+Sl5fH4YcfToMGDWjcuDFHHnkkRx555JZtcx06dOCwww6jXbt29OzZkyOOOIITTzyRSpUq7VYtkiRJ0r8pc8/oBAIBEmOjI/IJBAJF8jskJSVt9fP111/PoEGDuP/++xk1ahQTJ06kXbt2ZGdn7/Q+MTEx2/zd5OfnF0mN/1+FChUYP348AwcOpFatWtx+++106NCB9evXExUVxbBhwxg6dCitW7fmmWeeoUWLFsyfPz8stUiSJEllLuiUFrGxseTl5e3S2NGjR9O/f3/69u1Lu3btqFmz5pbneYpLq1atGD169DZ1NW/enKio0EpWdHQ0PXr04OGHH2by5MksWLCA77//HgiFrO7du3PXXXcxYcIEYmNjGTRoULH+DpIkSSo/ytzWtdKiYcOG/PrrryxYsIDk5OSdHhDQrFkzPv30U/r06UMgEOC2224L28rM6tWrmThx4lbf1apVi+uuu4699tqLe+65h1NOOYWff/6ZZ599lueffx6AL7/8knnz5nHggQdSqVIlvvrqK/Lz82nRogW//vor3333HUcccQTVq1fn119/ZfXq1bRq1Sosv4MkSZLkik6EXH/99URFRdG6dWuqVau20+dtHn/8cSpVqsR+++1Hnz596NmzJ507dw5LXe+99x6dOnXa6vPyyy/TuXNnPvzwQ95//33atm3L7bffzt13303//v0BqFixIp9++imHHnoorVq14sUXX2TgwIG0adOGlJQUfvzxR4466iiaN2/OrbfeymOPPUavXr3C8jtIkiRJgYJScMZveno6qamppKWlkZKSslXf5s2bmT9/Po0aNSI+Pj5CFaoo+Z+pJEmSdmRn2eCfXNGRJEmSVOYYdCRJkiSVOQYdSZIkSWWOQUeSJElSmWPQkSRJklTmGHQkSZIklTkGHUmSJElljkFHkiRJUplj0CmEgoICNmbl8ufG7EiXIkmSJGknDDqFkJmTx9zVGSxdn0l+QcEe3evggw/m6quvLprC/tK/f3+OO+64IhsnSZIklVYGnUJIiIkiJipI/l8rO5IkSZJKJoNOIQQCASrERwOQlpmz2/fp378/I0eO5KmnniIQCBAIBFiwYAEAU6dOpVevXiQnJ1OjRg3OPPNM1qxZs+Xajz/+mHbt2pGQkECVKlXo0aMHGzdu5M477+TNN99k8ODBW+45YsSI3apv5MiR7L333sTFxVGrVi1uvvlmcnP/F+x2VAPAiBEj2HvvvUlKSqJixYp0796dhQsX7vbflSRJkrQ7oiNdQJErKICcTWG7fWpUDn/mbCJjw2YKEvMJBAL/64xJhH/+vANPPfUUs2bNom3bttx9990AVKtWjfXr13PooYdy/vnn88QTT5CZmclNN93EySefzPfff8/y5cvp168fDz/8MH379mXDhg2MGjWKgoICrr/+eqZPn056ejqvv/46AJUrVy7077d06VKOOuoo+vfvz1tvvcWMGTO44IILiI+P584779xpDbm5uRx33HFccMEFDBw4kOzsbMaOHbv135EkSZJUDMpe0MnZBPfXDtvtKwDtdtT5n2UQm/Sv90hNTSU2NpbExERq1qy55ftnn32WTp06cf/992/57rXXXqNevXrMmjWLjIwMcnNzOf7442nQoAEA7dr9r5qEhASysrK2umdhPf/889SrV49nn32WQCBAy5YtWbZsGTfddBO33347y5cv32EN69atIy0tjd69e9OkSRMAWrVqtdu1SJIkSbvLrWslyKRJk/jhhx9ITk7e8mnZsiUAc+fOpUOHDhx22GG0a9eOk046iZdffpk///yzSGuYPn06++6771arMN27dycjI4MlS5bstIbKlSvTv39/evbsSZ8+fXjqqadYvnx5kdYnSZIk7Yqyt6ITkxhaWQmj9ZuyWfxnJnHRUTSvkbz13HsgIyODPn368NBDD23TV6tWLaKiohg2bBhjxozh22+/5ZlnnuGWW27h119/pVGjRns09676txpef/11rrzySr7++ms++OADbr31VoYNG0a3bt2KpT5JkiQJyuKKTiAQ2j4Wxk9ySirEJLE5EEdWIP5/fYV4FiU2Npa8vLytvuvcuTN//PEHDRs2pGnTplt9kpKS/vr1AnTv3p277rqLCRMmEBsby6BBg3Z4z8Jq1aoVP//8MwX/OD579OjRVKhQgbp16/5rDQCdOnViwIABjBkzhrZt2/Lee+/tUU2SJElSYZW9oFMMooNBkuKiAEjfvHunrzVs2JBff/2VBQsWsGbNGvLz87nssstYt24d/fr147fffmPu3Ll88803nHPOOeTl5fHrr79y//33M27cOBYtWsSnn37K6tWrtzwH07BhQyZPnszMmTNZs2YNOTk7ri0tLY2JEydu9Vm8eDGXXnopixcv5oorrmDGjBkMHjyYO+64g2uvvZZgMLjTGubPn8+AAQP4+eefWbhwId9++y2zZ8/2OR1JkiQVu7K3da2YpCTEkJGVS1pmLtUqFP7666+/nrPPPpvWrVuTmZnJ/PnzadiwIaNHj+amm27iiCOOICsriwYNGnDkkUcSDAZJSUnhxx9/5MknnyQ9PZ0GDRrw2GOP0atXLwAuuOACRowYQdeuXcnIyOCHH37g4IMP3u78I0aMoFOnTlt9d9555/HKK6/w1VdfccMNN9ChQwcqV67Meeedx6233hr6vXdSw8qVK5kxYwZvvvkma9eupVatWlx22WVcdNFFhf8LkiRJkvZAoOCfe5RKqPT0dFJTU0lLSyMlJWWrvs2bNzN//nwaNWpEfHx8sdWUnZvPjBXpALSqlUJMlItjRSVS/5lKkiSp5NtZNvgn/+l8N8VGB0mIDW1f27Cb29ckSZIkhYdBZw+kxMcAkJ6ZG+FKJEmSJP2TQWcPpCaEgs6GrFzy8kv8DkBJkiSp3DDo7IG46CCx0UEKCgrIcPuaJEmSVGIYdPZAIBD43/a1zW5fkyRJkkqKMhN0InV4XErC30Enh/ySf4BdqVAKDgKUJElSCVfqg05UVOjks+zs7IjMnxQbRXQwSF5+AZuyXNUpCps2bQIgJiYmwpVIkiSptCr1LwyNjo4mMTGR1atXExMTQzBY/NktIZhHenYO69IhGt/7srsKCgrYtGkTq1atomLFiltCrCRJklRYpT7oBAIBatWqxfz581m4cGFEasjMyWNtRjbrggE2php09lTFihWpWbNmpMuQJElSKVbqgw5AbGwszZo1i9j2tc3ZeVz3wmiyc/N58cwuNKteISJ1lAUxMTGu5EiSJGmPlYmgAxAMBomPj8xqSnw8NK1VmWHTVvLdrD9pV79aROqQJEmSFFLqDyMoKY5oXQOAb/9YGeFKJEmSJBl0ishhrWoQDMC05elMW5Ye6XIkSZKkcs2gU0QqJ8VyVLtaADz13awIVyNJkiSVbwadInTVYc0IBOCbP1byx7K0SJcjSZIklVsGnSLUrEYF+rSvDcCTw2dHuBpJkiSp/DLoFLErD2tGMADDpq1kyhJXdSRJkqRIMOgUsabVkzm2Yx0AnhzuszqSJElSJBh0wuCKQ5sSDMB3M1YxafH6SJcjSZIklTsGnTBoXC2Zvp3qAq7qSJIkSZFg0AmTKw9rSlQwwA8zVzN+0Z+RLkeSJEkqVww6YdKgShIndP77WR1PYJMkSZKKk0EnjK44tBnRwQA/zlrN7wvXRbocSZIkqdww6IRRvcqJnNgl9KzOE8Nc1ZEkSZKKi0EnzC47pCkxUQF+mrOG3xa4qiNJkiQVB4NOmNWrnMhJXesB8MQwT2CTJEmSioNBpxj8vaozZu5ahk1bGelyJEmSpDLPoFMM6lRM4Nz9GwHwn0FTWL8pO8IVSZIkSWWbQaeYXNOjOU2qJbF6QxZ3fTEt0uVIkiRJZZpBp5jEx0Tx6EkdCAZg0ISlbmGTJEmSwsigU4w61a/EBQc2BkJb2P7c6BY2SZIkKRwMOsXsmh7NaVo9mdUbsrjziz8iXY4kSZJUJhl0itk/t7ANnriMb/5YEemSJEmSpDLHoBMBHetV5KKDmgBwy6CpbmGTJEmSiphBJ0Ku7tGMZtWTWZORxR2fu4VNkiRJKkoGnQiJiw5tYYsKBvh80jK+nro80iVJkiRJZYZBJ4I61KvIRX+dwnbLoKmszciKcEWSJElS2WDQibCrejSjeY1k1m7M5rqPJpGfXxDpkiRJkqRSz6ATYXHRUTx1aifiooOMmLmal0fNi3RJkiRJUqln0CkBWtVK4Y4+bQB4+JuZ/L7wzwhXJEmSJJVuBp0Sot/e9ejToTZ5+QVcOXAC6zd55LQkSZK0uww6JUQgEOD+vm1pWCWRpeszuf6jyRQU+LyOJEmStDsMOiVIhfgYnj2tM7FRQYZPX8lroxdEuiRJkiSpVCp00Pnxxx/p06cPtWvXJhAI8Nlnn+3ytaNHjyY6OpqOHTsWdtpyo22dVG7t3QqAB4dOZ9Li9ZEtSJIkSSqFCh10Nm7cSIcOHXjuuecKdd369es566yzOOywwwo7ZblzZrcG9Gpbk5y8Ai4fOJ60zJxIlyRJkiSVKtGFvaBXr1706tWr0BNdfPHFnHbaaURFRRVqFag8CgQCPHhCe6YuS2Pxukxu/mQyz5/emUAgEOnSJEmSpFKhWJ7Ref3115k3bx533HHHLo3PysoiPT19q095k5oQw7P9OhMTFWDo1BW8OWZBpEuSJEmSSo2wB53Zs2dz880388477xAdvWsLSA888ACpqalbPvXq1QtzlSVTh3oVGdAr9LzOfV9NZ/wi368jSZIk7YqwBp28vDxOO+007rrrLpo3b77L1w0YMIC0tLQtn8WLF4exypLtnO4NOardX8/rvDuedRt9v44kSZL0bwr9jE5hbNiwgXHjxjFhwgQuv/xyAPLz8ykoKCA6Oppvv/2WQw89dJvr4uLiiIuLC2dppUYgEOChE9ozY/kG5q3ZyFXvT+CNc/YmKujzOpIkSdKOhHVFJyUlhSlTpjBx4sQtn4svvpgWLVowceJE9tlnn3BOX2ZUiI/h+TM6Ex8TZNTsNTzz/exIlyRJkiSVaIVe0cnIyGDOnDlbfp4/fz4TJ06kcuXK1K9fnwEDBrB06VLeeustgsEgbdu23er66tWrEx8fv8332rmWNVO4v287rv1wEk99N5tO9StxUPNqkS5LkiRJKpEKvaIzbtw4OnXqRKdOnQC49tpr6dSpE7fffjsAy5cvZ9GiRUVbpQA4vnNd+u1dn4ICuPr9CSxbnxnpkiRJkqQSKVBQUFAQ6SL+TXp6OqmpqaSlpZGSkhLZYqZ+AsFoaH1sRKbfnJPHiS+OYerSdDrWq8iHF+1LbHSxnBIuSZIkRdyuZgP/Cbkw5v4AH58Ln1wAi36JSAnxMVG8cHoXUuKjmbh4Pfd/NT0idUiSJEklmUGnMBodCC2OgrwsGHgqrInMoQD1KifyxCkdAXhjzAKGTlkekTokSZKkksqgUxjBKDjhVajTBTL/hHdOgIxVESnlsFY1uOTgJgDc8fkfbNicE5E6JEmSpJLIoFNYsYnQ7wOo1AjWL4R3T4KsjIiUcnWPZjSqmsSqDVk8McwjpyVJkqS/GXR2R3I1OOMTSKwCyyfCx+dAXm6xlxEXHcVdx7QB4M2fFzB9eXqx1yBJkiSVRAad3VWlSWhlJzoeZn8LQ66FCBxgd2DzahzVriZ5+QXc+tlU8vNL/CF6kiRJUtgZdPZEvb1Cz+wQgPFvwqhHI1LGbb1bkxgbxe8L/+Tj8UsiUoMkSZJUkhh09lSr3tDr4VD7+3th4sBiL6FWagJX92gGwINDZ7B+U3ax1yBJkiSVJAadorDPhbDflaH24Mtg3GvFXsI53RvRvEYy6zZm88g3M4t9fkmSJKkkMegUlR53QaczoCAPvrwGvr0N8vOLbfqYqCB3H9sWgPfGLmLS4vXFNrckSZJU0hh0ikowCMc8Cwf/J/TzmKfho7MhJ7PYSujWuAp9O9WhoABuGzyVPA8mkCRJUjll0ClKgQAcfBP0fQmiYmH65/BGb8hYXWwlDDiqJRXiopm8JI2BYxcV27ySJElSSWLQCYcOp8CZn0F8RVg6Dl45DFYXz3Mz1SvEc90RzQF45JuZrM3IKpZ5JUmSpJLEoBMuDbvD+cOhUkNYvxBePRzm/1gsU5/RrQGta6WQlpnDdR9NYnNOXrHMK0mSJJUUBp1wqtoMzv8O6u0Dm9Pg7eNh+hdhnzY6KsgDx7cjLjrIiJmrueCtcWRmG3YkSZJUfhh0wi2pKpz1ObQ+FvJz4MOzYeqnYZ+2Q72KvH7OXiTGRjFq9hrOfn0sGVm5YZ9XkiRJKgkMOsUhJh5OeA3anxo6fvqT82Dyh2Gfdr8mVXn7vL2pEBfN2PnrOOOVX0nblBP2eSVJkqRIM+gUl6hoOO556HgGFOTDpxfCxPfCPm2XBpV574JuVEyMYeLi9fR7+RcPKJAkSVKZZ9ApTsEoOOYZ6NIfKIDPLoXf3wj7tO3qpvL+hd2omhzLtOXpnPrSL6xK3xz2eSVJkqRIMegUt2AQej8Je18IFMAXV8HYl8M+bcuaKXxw0b7UTIln9qoMTnnpF5atL76XmUqSJEnFyaATCYEA9HoYul0W+vmr6+Hn58M+bZNqyXx40b7UrZTA/DUbuejt3ykoKAj7vJIkSVJxM+hESiAAPe+D7leHfv5mAHx/H4Q5eNSvksgHF+1LUmwUU5am8e20lWGdT5IkSYoEg04kBQLQ40446ObQzz8+DJ9dArnZYZ22TsUEzuneCIAnhs0iP99VHUmSJJUtBp1ICwTgkAGh53YCUTBpILx3UugFo2F0/gGNqBAXzYwVGxg6dUVY55IkSZKKm0GnpOh6DvR7H2KSYN4IeP0oSFsatukqJsZy3gGhVZ0nh88iz1UdSZIklSEGnZKk+RFwzhBIqg4rp8IrPWDF1LBNd+7+jUiJj2b2qgy+nLwsbPNIkiRJxc2gU9LU7gTnD4eqzWHDMni9V2iFJwxS4mO48MDGADw1fDa5eflhmUeSJEkqbgadkqhSAzjvW2jQHbLS4Z0TYNY3YZmqf/dGVEqMYd6ajQye6KqOJEmSygaDTkmVUAnOHARt+kJ+LnxyPqyZU+TTJMdFc9FBTQB46rvZ5LiqI0mSpDLAoFOSRcdB35egXrfQys4Hp0NWRpFPc9a+DaiSFMuidZv4dPySIr+/JEmSVNwMOiVddCyc/CYk14TVM2DwpUX+UtHE2GguOTi0qvP0d3PIznVVR5IkSaWbQac0qFATTn4LgjEwbTCMfqrIpzijWwOqVYhj6fpMPhy3uMjvL0mSJBUng05pUX8f6PVgqP3dXTD3+yK9fXxMFJf9tarz3A9z2JyTV6T3lyRJkoqTQac06XoedDwDCvLh4/Pgz4VFevtT965PrdR4lqdt5t1fFxXpvSVJkqTiZNApTQIBOPqx0Lt2MtfBB2dATmaR3T4+JorLDmkKwD1fTuPY50bz5pgFrNuYXWRzSJIkScUhUFBQxE+2h0F6ejqpqamkpaWRkpIS6XIib/1ieOkg2LQWOvSD414IhaAikJ2bz82fTGbwpGXk5Yf+qxEdDHBwi+qc0LkOh7aqTlx0VJHMJUmSJBXWrmYDg05pNf9HeOs4KMiD3k9A13OL9PZrMrL4fOIyBk1YypSlaVu+T4mP5sIDG3P5oc2KdD5JkiRpVxh0yoPRT8Ow2yAmCS4dA5UahmWa2Ss38OmEpXw2YSnL0zYD8P6F3ejWuEpY5pMkSZJ2ZFezgc/olGb7Xg4N9oecjTD4csgPz/tvmtWowE1HtuSnmw7l5K51AXj6u9lhmUuSJEkqCgad0iwYhGOfDa3oLBgFv70S1umiggGu6tGcmKgAY+au5bcF68I6nyRJkrS7DDqlXeVGcPhdofbwO2DdvLBOV6diAid2cVVHkiRJJZtBpyzoeh40PAByNoV1C9vfLj24KdHBAKNmr+H3hX+GdS5JkiRpdxh0yoJ/bmFbOBrGvhTW6epVTuSEzq7qSJIkqeQy6JQVlRrCEXeH2sPvhLVzwzrdZYc0JSoYYOSs1UxcvD6sc0mSJEmFZdApS7qcC40OhNxMGHxZWLew1a+SSN9OdQBXdSRJklTyGHTKkmAQjnkWYpNh0c/w64thne6yQ5oSDMD3M1Yxecn6sM4lSZIkFYZBp6yp1ACOuCfU/u7usG5ha1Q1ieM6uqojSZKkksegUxZ1OQcaHxzawvbZJZCfF7apLjs0tKozfPoqpi5NC9s8kiRJUmEYdMqiQOCvLWwVYPGv8MsLYZuqSbVk+nSoDbiqI0mSpJLDoFNWVawHPe8Ntb+/B9aEL4RccWhTAgH4dtpKpi1LD9s8kiRJ0q4y6JRlnc+GJodC7uawbmFrWr0CvduHVnWe+d5VHUmSJEWeQacsCwTgmGcgLgWW/AY/Pxu2qf5e1Rk6dQXjF/0ZtnkkSZKkXWHQKetS60LP+0Pt7++D1TPDMk3zGhU4oXNdAG7+ZDLZueF7h48kSZL0bww65UGnM6Dp4ZCXFdrClpcblmluOaoVVZJimbUygxdHhu9Ya0mSJOnfGHTKg0AAjnka4lJh6e/w8zNhmaZSUix3HNMGgGe/n8OcVRvCMo8kSZL0bww65UVKbej1YKj9w/2wanpYpunTvhaHtqxOdl4+N38yhfz8grDMI0mSJO2MQac86dAPmh8Jedlh28IWCAS457i2JMVGMW7hn7w7dlGRzyFJkiT9G4NOeRIIQO8nIT4Vlk2AHx8OyzR1KiZw45EtAXho6AyWp2WGZR5JkiRpRww65U1KLej1SKg98iEY/XRYpjmjWwM6169IRlYut302lYICt7BJkiSp+Bh0yqP2J8OBN4baw26DUY8V+RRRwQAPntCemKgAw6ev4qspK4p8DkmSJGlHDDrlUSAAh94CB/8n9PN3d8PIot/G1rxGBS49uCkAd3w+lfWbsot8DkmSJGl7DDrl2cE3wWG3h9o/3Bc6ja2It5hdekgTmlZPZk1GNvd/FZ6T3iRJkqT/z6BT3h1wHRx+T6g98qHQ6k4Rhp246CgeOqEdgQB8OG4Jo2avLrJ7S5IkSTti0BF0vxJ6PhBq//Q4DLu9SMNOlwaVOXvfhgDc+PFk0jfnFNm9JUmSpO0x6Chk30v/dxrbmKfh8ysg888iu/2NR7agQZVElqdt5p4vphXZfSVJkqTtMejof/a5EI5+PNSe8DY83Ql+eQFy9/wQgcTYaB49qQOBAHz0+xK+n7Fyj+8pSZIk7YhBR1vb6zw48zOo1iq0ovP1zfB8N5gxZI+3s+3VsDLndW8EwM2fTPEUNkmSJIWNQUfbanIIXPwT9H4SkqrBurnw/mnwRm9YNnGPbn19zxY0qZbEqg1Z3Pn5H0VSriRJkvT/GXS0fVHR0PUcuHJC6GS26HhY+BO8dDB8deNur+7Ex0Tx6EkdCAbgs4nL+HqqLxKVJElS0TPoaOfiKoTetXP5OGh3MlAAY/8Ls77Z7Vt2ql+Jiw5qAsAtg6awNiOriIqVJEmSQgw62jUV68EJL0P3q0I/D78T8vN2+3ZX92hGixoVWLsxm9sGT6WgiF9UKkmSpPLNoKPC2f8aiK8Iq6fD5A92+zZx0VE8dnIHooIBvpqygi8mLy+6GiVJklTuGXRUOAmV4IBrQ+3v74Oczbt9q7Z1UrnskKYA3D54KivTd/9ekiRJ0j8ZdFR4e18IKXUgfQn89vIe3eryQ5rSpnYK6zflcO4bv5GRlVtERUqSJKk8M+io8GIS4JD/hNqjHoPM9bt9q9joIC+c3oWqybH8sSydS98dT05eftHUKUmSpHLLoKPd06Hf/14qOvqpPbpV/SqJvHr2XiTERPHjrNX859MpHk4gSZKkPWLQ0e4JRoWOnQb45QVIX7ZHt+tQryLPnd6JYAA++n0JTw6fXQRFSpIkqbwy6Gj3tegF9bpBbiaMeHCPb3doyxrce1w7AJ76bjbvj120x/eUJElS+WTQ0e4LBODwu0LtCe/A6ll7fMvT9qnPFYeGTmK75bOp/DBj1R7fU5IkSeWPQUd7pn43aHE0FOTB93cXyS2vPbw5x3euQ15+AZe+O57JS9Zv6duYlcu0Zel8PXU5/x05l/uGTGP+mo1FMq8kSZLKjkBBKXjqOz09ndTUVNLS0khJSYl0Ofr/Vs2AF/aFgnw4bzjU22uPb5mdm895b/7GqNlrqJwUS8MqiSxat4k1GdnbjG1dK4UvrtifqGBgj+eVJElSybar2cAVHe256i2h42mh9vA7oAiyc2x0kOdP70zrWims25jN+EXrt4SciokxdKibSp8OtUmJj2ba8nTe83keSZIk/YMrOioaaUvgmS6Quxl6PwldzymS267bmM3nE5dStUIcDSonUb9KIqkJMVv63/p5AbcP/oPUhBh+uP5gKifFFsm8kiRJKplc0VHxSq0Lh9wSan99M6z8o0huWzkplv7dG9G7fW3a1U3dKuQAnLZ3fVrVSiEtM4dHvplZJHNKkiSp9DPoqOjsezk0PTy0qvNRf8gO/yEB0VFB7j62DQDv/7Zoq4MLJEmSVH4ZdFR0gkHo+yJUqAVrZsFXNxTLtHs1rEzfTnUoKIDbB/9Bfn6J340pSZKkMDPoqGglVYUTXoFAECa+C5PeL5ZpB/RqSVJsFBMXr+fj8UuKZU5JkiSVXAYdFb2G+8NBN4XaX14La2aHfcrqKfFc3aM5AA8NnUFaZk7Y55QkSVLJZdBReBx4AzQ8AHI2wkfnQM7msE959n4NaVItibUbs3li2KywzydJkqSSy6Cj8AhGwfEvQ2JVWDkFvr0l7FPGRge585jQwQRv/byA6cvTwz6nJEmSSiaDjsInpRb0/W+o/dsrMG1w2Kc8oFk1erWtSX4B3PH5H5SC10RJkiQpDAw6Cq9mPaD7VaH24Ctg/eKwT3nL0a2Ijwkydv46Pp+0LOzzSZIkqeQx6Cj8Dr0N6nSFrLTQy0TDrG6lRC47uCkAD3w1g03ZuWGfU5IkSSWLQUfhFxUDxzwDwWiY8SXM+CrsU15wYGPqVExgRfpmXvpxXtjnkyRJUsli0FHxqNEa9r0s1B56I2Rv3LXrcjbDvBGQV7hVmfiYKAYc1RKA/46cx/K0zEJdL0mSpNLNoKPic9BNkFoP0hbDyIf/fXzOZni7L7x1LHxyLhTyYIGj29Wia4NKZObk8cjXM3ezaEmSJJVGBh0Vn9gkOOqRUPvnZ2HltB2Pzc+HQRfCojGhn6cNhp8eL9R0gUCA23q3BuDTCUuZtHj9bhQtSZKk0sigo+LVohe07A35uTDk2lCg+f8KCuCb/4TCTTAGup4X+v67e2D2sEJN16FeRY7vVAeAe76c5nHTkiRJ5YRBR8XvyAchJgkW/QwT3922/+dn4dcXQu2+L0Lvx6HLOUABfHwerJ1bqOluOLIFCTFRjFv4J0OmLN/z+iVJklTiGXRU/CrWg0MGhNrDboONa//XN+Vj+PbWUPuIe6HdiaF2r4eh3j6hI6rfPw2yNuzydLVSE7jooMZA6LjpzTl5RfFbSJIkqQQz6Cgy9rkYqreBzD9h+O2h7+aNhEEX/9V/Cex7+f/GR8fCyW9BhVqwekZo3Pa2ve3ARQc2oVZqPEvXZ/LqT/OL8BeRJElSSWTQUWRExUDvJ0LtCe/Ab6/AB2dAfg60Pg563g+BwNbXVKgJp7wDUbGh9/GMemyXp0uIjeLGI1sA8PwPc1i1YXMR/SKSJEkqiQw6ipz6+0Dns0PtIddBVjo06A59/wvBHfxXs25XOPqv09d+uA9mfr3L0x3boQ4d6lVkY3Yej387aw+LlyRJUklm0FFk9bgTEquE2tVawqnvQkz8zq/pfCbsdT5QAJ9eAGtm79JUwWCA23u3AuCDcYv5Y1na7tctSZKkEs2go8hKrAz93g8FlzM+hYRKu3Zdzweg/n6hVaBvb9vl6bo0qEyfDrUpKID7hkz3uGlJkqQyyqCjyKu3Nxz9GKTW2fVromPhmKdD7dnfwPrFu3zpTUe2IDYqyJi5axk1e00hi5UkSVJpYNBR6VW1GTQ6EAry4fc3dvmyupUSOXPfBgA8/M0M8vNd1ZEkSSprDDoq3bqeF/pz/FuQm73Ll116cBOS46KZujSdr6b6ElFJkqSyxqCj0q3l0ZBcEzauCh05vYuqJMdxwQGhl4g+9u0scvJ2/Z08kiRJKvkMOirdomKg81mh9rjXCnXpeQc0okpSLPPXbOSjcUvCUJwkSZIixaCj0q/L2RAIwoJRsHrmLl+WHBfN5Yc2BeCp72aRmZ0XrgolSZJUzAw6Kv1S60LzI0PtQq7qnLZPfepUTGBlehZv/ryg6GuTJElSRBh0VDb8fSjBxIGQvXGXL4uLjuLaw5sD8PwPc0jblBOO6iRJklTMDDoqG5ocCpUaQlYaTP20UJce16kOzWskk745l//+ODc89UmSJKlYGXRUNgSD0OWcUHvcq4W6NCoY4IaeLQF4bfR8VqVvLurqJEmSVMwMOio7Op0BUbGwbAIsHV+oS3u0qk7n+hXZnJPPM9/PCVOBkiRJKi4GHZUdSVWh9XGhdiFXdQKBADcdGVrVGTh2EQvX7vpzPpIkSSp5DDoqW/b661CCKZ9A5p+FunSfxlU4uEU1cvMLuOPzP1iTkRWGAiVJklQcCh10fvzxR/r06UPt2rUJBAJ89tlnOx3/6aefcvjhh1OtWjVSUlLYd999+eabb3a3Xmnn6u0D1dtAbiZMer/Ql9/QswXBAIyYuZruD37P7YOnsnjdpjAUKkmSpHAqdNDZuHEjHTp04Lnnntul8T/++COHH344X331Fb///juHHHIIffr0YcKECYUuVvpXgQDsdW6oPe41KCgo1OVtaqfyxjl706FuKlm5+bz180IOfnQE13wwkZkrNoShYEmSJIVDoKCgkP8k+M+LAwEGDRrEcccdV6jr2rRpwymnnMLtt9++S+PT09NJTU0lLS2NlJSU3ahU5UrWBnisJWRnwNlfQKMDC32LgoICfp63lhdGzGXU7DVbvj+sZXWu7tGcdnVTi7JiSZIk7aJdzQbF/oxOfn4+GzZsoHLlyjsck5WVRXp6+lYfaZfFVYD2J4faIx+G/LxC3yIQCLBfk6q8fd4+fHH5/hzdrhaBAHw3YxUnvDiGBWs8rECSJKkkK/ag8+ijj5KRkcHJJ5+8wzEPPPAAqampWz716tUrxgpVJnS7FKITYMEoGLZrK4c70q5uKs+d3pnvrj2ILg0qkZ2bzyPfziyiQiVJkhQOxRp03nvvPe666y4+/PBDqlevvsNxAwYMIC0tbctn8eLFxVilyoSqzaDvC6H2z8/ChHf2+JaNqyVz73FtCQRgyOTlTFq8fo/vKUmSpPAotqDz/vvvc/755/Phhx/So0ePnY6Ni4sjJSVlq49UaG36wkE3h9pfXA2LftnjW7aqlULfTnUAeGDodPbgETdJkiSFUbEEnYEDB3LOOecwcOBAjj766OKYUgo56CZodQzk58D7p8P6RXt8y+uOaEFsdJBf5q1jxKzVRVCkJEmSilqhg05GRgYTJ05k4sSJAMyfP5+JEyeyaFHoHyAHDBjAWWedtWX8e++9x1lnncVjjz3GPvvsw4oVK1ixYgVpaWlF8xtIOxMMQt8XoWY72LQGBp4GWRl7dMs6FRPov19DAB4aOoO8fFd1JEmSSppCB51x48bRqVMnOnXqBMC1115Lp06dthwVvXz58i2hB+Cll14iNzeXyy67jFq1am35XHXVVUX0K0j/IjYJTh0ISdVg5RQYdBHk5+/RLS89uAkp8dHMWLGBzyYsLaJCJUmSVFT26D06xcX36KhILPoV3uwNedlw4A1w6K17dLsXR87lwaEzqFMxge+uO4j4mKgiKlSSJEk7UmLfoyNFTP19oM9TofaPj8CUj/fodv33a0it1HiWrs/k7Z8XFkGBkiRJKioGHZUvHU+D/a4ItT+/AlbN2O1bxcdEcc3hzQF49oc5pG3KKYoKJUmSVAQMOip/etwFjQ+GnE3w0dmQvXG3b3VC57o0r5FMWmYOz4+cU3Q1SpIkaY8YdFT+BKPg+JchuQasngFDrt/tW0UFA9x0ZEsAXh+9gGXrM4uqSkmSJO0Bg47Kp+TqcOJrEAjCpPdgwju7fatDW1Zn70aVyc7N54lhs4qwSEmSJO0ug47Kr4b7wyH/CbWHXA8rp+3WbQKBAAN6hVZ1Pvp9Ccc+N5rXR89n1YbNRVWpJEmSCsnjpVW+5efDuyfA3O+hanO44AeIS96tWz309Qz+O3Iuf78/NBiA7k2rcmzHOvRsU4MK8TFFWLgkSVL5tKvZwKAjbVwDL+4PG5ZD+1Og738hENitW63ekMWQycsYPGkZExat3/J9bHSQo9rW5O7j2pJi4JEkSdptBh2pMBaOgTd6Q0Ee9Hkaupy957dcu5HPJy7js4lLmbs6dLJb7/a1eKZfJwK7GaQkSZLKO18YKhVGg/3g0FtD7aE3woqphb/HxrVbvZenQZUkrjisGcOvPYh3ztuHqGCALycvZ9CEpUVUtCRJknbEoCP9rfvV0OwIyN0MH54ZCi67Km0pvNgdXtgPlv6+VVcgEGD/ZlW5+rBmANw++A8Wr9tUhIVLkiTp/zPoSH8LBkPP56TWh3XzYOCpkL0LgSRrA7x3cugZn4I8+P6+7Q679JCm7NWwEhlZuVz9wURy8/KL+BeQJEnS3ww60j8lVoYzPob4VFgyFj69APLzdjw+Lxc+6g8rp0JiVQhGw9zvYNEv2wyNCgZ4/OSOVIiL5veFf/L8iLnh+z0kSZLKOYOO9P9VawH93oeoOJjxJQy9CbZ3ZkdBAQy9AeYMh+gEOP1D6HhaqO+H+7d763qVE7n7uDYAPPXdbCYs+jNcv4UkSVK5ZtCRtqfBfnD8S0AAfnsZRj+17Zgxz8C410JjTngF6nSBA2+AYAzMHwkLftrurY/rWIc+HWqTl1/A1R9MJCMrN6y/iiRJUnlk0JF2pM1x0POvlZnhd8Dkj/7XN20wDLst1O55H7TqHWpXrA+dzwq1f7h/uytBgUCAe49rS52KCSxcu4m7v/gjfL+DJElSOWXQkXZm30th38tD7c8ugXkjYck4+PTC0Hd7XQDdLt36mgOuC217Wzg6tLKzHakJMTx2cgcCAfhw3BKGTlkexl9CkiSp/DHoSP/m8HugTV/Iz4EPzoD3TgkdQd2sJxz5IPz/l3+m1oGu54TaO1jVAejWuAoXH9QEgJs/ncKKtM3h/C0kSZLKFYOO9G+CQTjuRWjQHbLSYdMaqNkeTnwNoqK3f83+10B0PCz+FeZ8t8NbX9OjOW3rpJCWmcP1H00iP3/7oUiSJEmFY9CRdkVMPJz6LtTbB6q3gdM+gLjkHY+vUBP2Oj/U/uG+Ha7qxEYHeerUTsTHBPlpzhpeH7Og6GuXJEkqhww60q5KqATnfgOXjIaU2v8+vvvVEJMIy8bDrG92OKxJtWRuPbo1AA99PYMZK9KLqGBJkqTyy6AjFUYgsO0zOTuSXA32/uvQgp2s6gCcvk99Dm1ZnezcfK5+fyKbc3byklJJkiT9K4OOFE77XQmxybBicujlozsQCAR46IT2VEmKZcaKDTz6zcxiLFKSJKnsMehI4ZRUBfa5ONT+4QHIz9/h0GoV4nj4xPYAvPLTfEbPWVMcFUqSJJVJBh0p3Pa7HOJSYNUf8MenOx16WKsanL5PfQCu+3AS6zdlF0eFkiRJZY5BRwq3hEqhLWwA390NuVk7HX7L0a1oXDWJFembuWXQVAp28myPJEmSts+gIxWHfS+F5BqwfiGMe22nQxNjo3ny1I5EBwMMmbKcQROWFlORkiRJZYdBRyoOsUlw8IBQe+TDsDltp8Pb163I1T2aAXD74D9YvG5TuCuUJEkqUww6UnHpdCZUbQ6Z6+CnJ/91+CUHN6Vrg0pkZOVy86eT3cImSZJUCAYdqbhERUOPO0PtX16A9GU7Hx4M8NjJHYiPCTJ6zlo+Ge8WNkmSpF1l0JGKU4ujoF43yM2EH+7/1+ENqiRxdY/mANw7ZBprMnZ+kIEkSZJCDDpScQoE4Ih7Qu2J78Kq6f96yfn7N6J1rRTWb8rhni+nhblASZKkssGgIxW3entDqz5QkA/D7/zX4dFRQR46oT3BAAyeuIwfZq4Kf42SJEmlnEFHioTD7oBAFMz6GhaM/tfh7eqmcm73RgDcOmgqG7Nyw12hJElSqWbQkSKhajPo0j/UHnYb7MKJatce0Zy6lRJYuj6Tx76dFd76JEmSSjmDjhQpB98MMUmw9HeY9tm/Dk+Mjea+vu0AeGPMfCYtXh/e+iRJkkoxg44UKcnVYb8rQu3v7oa8nH+95KDm1TiuY23yC+CmTyaTk5cf5iIlSZJKJ4OOFEn7XQ5J1WHdPHjlMJj1zb9uY7utd2sqJcYwY8UGXh41r5gKlSRJKl0MOlIkxVWA3k9AbDIsnwTvnQyvHg5zf9hh4KmSHMetR7cG4Knhs1mwZmNxVixJklQqGHSkSGvVG66aDPtdCdEJsOQ3ePs4eOPoHZ7IdnznOhzQrCpZuflc8u540jf/+7Y3SZKk8sSgI5UESVVCLxK9ahLsczFExcLC0fDGUfDWcZC+fKvhgUCAB45vR9XkOKYvT+f8N8exOScvMrVLkiSVQAYdqSSpUAN6PQRXToAu50AwGub9AN8M2GZo3UqJvHnuXlSIi2bs/HVc/t54cj2cQJIkCTDoSCVTal3o8yScMzT08/QvIGPVNsPa1E7llbO7EhcdZPj0Vdz0yRTy8//9nTySJEllnUFHKsnq7Q11ukJ+Lox/a7tD9mlchedO60xUMMAn45dw/1fTKdiFF5BKkiSVZQYdqaTb67zQn7+/Cfnbfw6nR+saPHxCewBe+Wk+L4ycW1zVSZIklUgGHamka9MX4itC2iKYM3yHw07oUpdbj24FwMNfz2Tg2EXFVKAkSVLJY9CRSrqYBOh4eqj926s7HXr+AY259OAmANwyaApfT12+0/GSJElllUFHKg26nhv6c/a3sH7nKzU39GxBv73rkV8AV38wkenL04uhQEmSpJLFoCOVBlWbQqODgAL4/Y2dDg0EAtx7XDsOaFaVzTn5XPzO76Rt8oWikiSpfDHoSKXF36s649+G3OydDo0KBnj61E7UqZjAwrWbuPqDCR47LUmSyhWDjlRatDwakmvAxlUw48t/HV4pKZb/ntmFuOggP8xczdPfzy6GIiVJkkoGg45UWkTFQOezQu1xr+3SJW3rpHJf33YAPDl8Nt9NXxmu6iRJkkoUg45UmnTpD4EgLBgFq2ft0iUndqnLmd0aAKHDCRas2RjGAiVJkkoGg45UmqTWheZHhtq7uKoDcFvv1nSuX5ENm3O56O3f2ZSdG6YCJUmSSgaDjlTa/H0owaT3IHvTLl0SGx3khTO6UDU5jpkrN3DTJ1MoKPBwAkmSVHYZdKTSpslhULE+bE6DPz7d5ctqpMTz/OmdiQ4G+GLSMl4bvSB8NUqSJEWYQUcqbYJB6HJOqP3bq4W6dO9Glbn16FYAPPbtTDZmuYVNkiSVTQYdqTTqdCYEY2DZeFg2oVCXnr1fQxpVTWJTdh5DpiwPU4GSJEmRZdCRSqPkatD62FD7+3shb9dXZgKBACd2qQvAx+OWhKM6SZKkiDPoSKVV96sgKg7mDIcvroJCHC5wQue6BAMwdsE6j5uWJEllkkFHKq1qtYeTXg+9V2fiOzD8jl2+tGZqPAc0qwbAx7+7qiNJksoeg45UmrU8Gvo8HWqPfgpGP73Ll57UNbR97ZPxS8jL96hpSZJUthh0pNKu85nQ465Qe9htMOHdXbrs8NY1qJgYw/K0zfw0Z00YC5QkSSp+Bh2pLNj/atjvilD78ytg5tB/vSQuOopjO9QG4KNxi8NYnCRJUvEz6EhlxeH3QIfToCAPPuoPC8f86yUnda0HwLfTVpK2KSfMBUqSJBUfg45UVgQCcMwz0LwX5G6G906FFVN2ekmb2im0rFmB7Nx8Pp+0tJgKlSRJCj+DjlSWREWHTmKrvy9kpcHH50F+3g6HBwKBLas6H3n6miRJKkMMOlJZE5MAp74H8amwZiZM+2ynw4/rWJuYqACTl6QxY0V68dQoSZIUZgYdqSxKrAzdLg21Rz4C+fk7HFolOY7DWtYA4KNxrupIkqSywaAjlVX7XAxxKbB6Okz/fKdD/36nzmcTlpKTt+NQJEmSVFoYdKSyKqFiKOwA/LjzVZ2DmlejWoU41m7M5vsZq4qnPkmSpDAy6EhlWbdLILYCrJwKM4fscFh0VJDjO9cB3L4mSZLKBoOOVJYlVoZ9Lgy1Rz4EBQU7HHpSl9Dpaz/MXMWqDZuLozpJkqSwMehIZd2+l0NscuidOjOH7nBY0+rJdKpfkbz8Aj6b4Dt1JElS6WbQkcq6xMqw9wWh9sgHd2lV54PfFpOXv+NxkiRJJZ1BRyoP9r0cYhJh+SSY9c0Oh/XuUIsKcdHMXb2RF0fOLcYCJUmSipZBRyoPkqrCXueH2jt5ViclPoY7j2kDwBPDZjFx8fpiKlCSJKloGXSk8mK/KyE6AZaNhznDdzjs+M516N2+Frn5BVz9/gQ2ZuUWY5GSJElFw6AjlRfJ1WCv80LtETt+VicQCHDfce2onRrPgrWbuOuLP4qxSEmSpKJh0JHKk/2uhOh4WDoO5n6/w2GpiTE8cUpHAgH4cNwSvpqyvBiLlCRJ2nMGHak8qVADup4bag+/E3Iydzh0n8ZVuPTgJgAM+HQKy9N2PFaSJKmkMehI5U33qyAuFVZMho/Pg7wdP4NzdY/mtK+bSlpmDtd+MIl8j5yWJEmlhEFHKm8q1IR+AyEqDmYOgSHX7vB5nZioIE+d2omEmCh+nreWl0fNK+ZiJUmSdo9BRyqPGnaHE16BQBDGvwkjHtjh0EZVk7jzmNYAPPrtTKYuTSuuKiVJknabQUcqr1ofA0c9GmqPfAh+e3WHQ0/uWo8j29QkJ6+AKz1yWpIklQIGHak82+s8OOjmUPur62H6F9sdFggEeOD4dtRMiWfe6o3c+PFkCnaw3U2SJKkkMOhI5d3BN0OX/lCQHzqcYMHo7Q6rlBTLc6d3JiYqwJApy31eR5IklWgGHam8CwTg6MehZW/Iy4KB/WDl9l8S2qVBJW7vHXpe58GhMxgzZ01xVipJkrTLDDqSIBgVOpyg/n6QlQbvngSb1m136BndGnBC57rkF8DlAyewbL3v15EkSSWPQUdSSEwC9HsPqjSF9KXwxVXbPXY6EAhwX9+2tKmdwrqN2Vzyzu9szsmLQMGSJEk7ZtCR9D8JlUIrO8FomP45THxvu8PiY6J48YwuVEyMYdKSNO78fPtb3SRJkiLFoCNpa7U7wSG3hNpDb4R187c7rF7lRJ4+tROBALz/22IGjl1UjEVKkiTtnEFH0ra6XwUNukN2Bnx6IeRt/705BzavxvVHtADgjsF/MHHx+mIsUpIkaccMOpK2FYyCvi9CXAosGQujHtvh0EsPbkLPNjXIzsvnknd+Z93G7GIsVJIkafsMOpK2r2J9OPqvgDPyIVgybrvDAoEAj57UgcbVklietplbP5viy0QlSVLEGXQk7Vj7k6HtiVCQB59eAFkZ2x1WIT6Gp0/tRHQwwFdTVvD5pGXFXKgkSdLWDDqSdu7oxyClLqybB98M2OGwtnVSueLQZgDc9tlUVqRtLq4KJUmStmHQkbRzCRVDz+sQgPFvwfQvdjj00kOa0KFuKumbc7nxk8luYZMkSRFj0JH07xodAN2vDLU/vxI2rt3usJioII+d3JG46CA/zlrNu7965LQkSYoMg46kXXPIrVCjLWSugx/u2+GwptWTuenIlgDcN2Q6C9Zs3OHYzTl5PPrNTI54YiSj56wp8pIlSVL5ZdCRtGuiY+HIB0Pt31+HldN2OLT/fg3Zt3EVMnPyuP6jSeTlb7uFbcycNRz55I88+8McZq3M4MqBE1i1wed6JElS0TDoSNp1jQ6AVn2gID90MMEOnsEJBgM8clJ7kuOiGbfwT14eNW9L37qN2Vz34SROe+VXFqzdRI2UOBpXTWLtxmxu+Ggy+dsJRZIkSYVl0JFUOIffA1GxMG8EzPp6h8PqVkrk9j6tAXj821lMX57OoAlL6PH4SD4Zv4RAAM7atwHDrz2IF8/sQlx0kJGzVvPGmAXF83tIkqQyzaAjqXAqN4Jul4ba39wCudk7HHpSl7r0aFWd7Lx8jn9+DNd8MIl1G7NpUaMCn1yyH3cf25YK8TE0r1GBW45uBcCDQ2cwfXl6cfwmkiSpDDPoSCq8A6+HpOqwbi6MfWmHwwKBAA8c357KSbFk5uQRFx3kxiNb8OWV+9O5fqWtxp7ZrQGHtQyFoisHTmBzTl64fwtJklSGGXQkFV5cBTjs9lB75MOwcccnplWrEMerZ3flvP0b8c3VB3LpwU2Jidr2f3oCgQAPn9ieahXimL0qg/uGTA9X9ZIkqRww6EjaPR1Pg5rtISttp8dNA3SqX4nberemYdWkbTtzMmHmUMjLpUpyHI+d1AGAt39ZyPBpK8NRuSRJKgcMOpJ2TzDqH8dNvwErpu7efb75Dww8FcY8BcCBzatx/v6NALjxk8msSvfIaUmSVHgGHUm7r2F3aH3sX8dN/2eHx03vUPYmmPxRqD3p/S3X33BkC1rXSgkdRf3RJI+cliRJhWbQkbRnDr8bouJg/kiY+VXhrp0xBLI3hNprZsHKPwCIi47i6X4diY8JMmr2Gt75dWERFy1Jkso6g46kPVOpIex7Waj9zS2Qm7Xr1056L/RnMDr05x+fbulqWr0CNx/ZEoBnv59DVq6nsEmSpF1n0JG05w64FpJrwJ/zQ1vQdkX6stBLRwEOHhD6c+qnW21/O22fBtRKjWfVhiw++X1p0dYsSZLKNIOOpD0XVwG6XxVqj3ka8vP//ZrJH4Se7am/H+xzMUQnhILSsglbhsRGB7nwwMYAvDhyLrl5u3BfSZIkDDqSikrnsyA+FdbO+fdndQoKYOLAULvDqRCXDC2ODP38j+1rAKfuVZ/KSbEsWreJIVOWh6FwSZJUFhl0JBWNuAqw1/mh9ugnd34C27LxsGYmRMdDm+NC37U5PvTnH59tdW1CbBTndm8IwPM/zPUENkmStEsMOpKKzj4Xh05gW/IbLPp5x+P+fo6nZe/QKhBAs8MhNhnSFoeu/4cz921Iclw0M1du4PsZq8JUvCRJKksMOpKKTnJ16HhaqD36qe2Pyc2GKR+H2h37/e/7mARocVSoPXXr7WupCTGcuW8DAJ79YQ4FhX1fjyRJKncMOpKK1n5XAAGY9TWsmr5t/+xvIHMdJNeExods3df27+1rgyB/6+Okz+3eiLjoIBMXr+fneWvDU7skSSozCh10fvzxR/r06UPt2rUJBAJ89tln/3rNiBEj6Ny5M3FxcTRt2pQ33nhjN0qVVCpUaQKt+oTaY57Ztv/vQwjanwzBqK37mhwa2sqWsWKbrW/VKsRxyl71AHhhxNyirlqSJJUxhQ46GzdupEOHDjz33HO7NH7+/PkcffTRHHLIIUycOJGrr76a888/n2+++abQxUoqJf4+anryh5D2j/ffbFwTWtGB/21x+6foOGj5V0j6f9vXAC48sDHRwQCjZq9h0uL1RVuzJEkqUwoddHr16sW9995L3759d2n8iy++SKNGjXjsscdo1aoVl19+OSeeeCJPPPFEoYuVVErU7QoN9of8HPj1hf99P+VjyM+FWh2heqvtX9v2r/9tmTYY8nK3vm2lRI7tWAeA50fMCUPhkiSprAj7Mzo///wzPXr02Oq7nj178vPPOzmRSVLp9/eqzrg3IHN9qD3pr21r21vN+VujgyChMmxaAwtGbdN9ycGNCQTgmz9WMnvlhiItWZIklR1hDzorVqygRo0aW31Xo0YN0tPTyczM3O41WVlZpKenb/WRVMo0Oxyqt4bsDTDutdDBBMsnQjAG2p644+uiYqD1MaH21E+26W5avQI9W9cE4IWRPqsjSZK2r0SeuvbAAw+Qmpq65VOvXr1IlySpsAIB2O/KUPvXF+H3N0Lt5j0hqcrOr217QujP6V+EjqP+fy49pAkAgycuY/G6TUVUsCRJKkvCHnRq1qzJypUrt/pu5cqVpKSkkJCQsN1rBgwYQFpa2pbP4sWLw12mpHBodyKk1IWMlfDrf0PfdTj1369r0B2Sa8Dm9TBvxDbd7etW5IBmVcnLL+Chr2f4Xh1JkrSNsAedfffdl++++26r74YNG8a+++67w2vi4uJISUnZ6iOpFIqKgX0v/euHgtCzN816/vt1wShofWyo/ce2p68BXN2jGcEAfDl5OfcOmW7YkSRJWyl00MnIyGDixIlMnDgRCB0fPXHiRBYtWgSEVmPOOuusLeMvvvhi5s2bx4033siMGTN4/vnn+fDDD7nmmmuK5jeQVLJ1Piv0bhwIrfBEx+7adW3+ennojCGQs3mb7i4NKvPwiR0AePWn+TzzvaewSZKk/yl00Bk3bhydOnWiU6dOAFx77bV06tSJ22+/HYDly5dvCT0AjRo1YsiQIQwbNowOHTrw2GOP8corr9Cz5y78W11JpV9cBej5ANTpAvtetuvX1dsHUupAVjrMGb7dISd2qcsdfVoD8PiwWbz20/yiqFiSJJUBgYJSsN8jPT2d1NRU0tLS3MYmlSff3AI/PwvVWsJ5wyB++//3//R3s3l82CwAHjmxPSd19QATSZLKql3NBiXy1DVJAkKntlWoBatnwCfnQ37edoddcWhTzt+/EQA3fTKZr6cuL84qJUlSCWTQkVRyVagBp74H0fEw+xsYfud2hwUCAW45uhWndK1HfgFcMXACP85aXby1SpKkEsWgI6lkq9MZjns+1B7zNEx8b7vDAoEA9x/fjqPb1SInr4AL3x7HuAXrirFQSZJUkhh0JJV8bU+AA28Mtb+4Chb9ut1hUcEAT5zSkYOaV2NzTj4Xvv07y9ZnFmOhkiSppDDoSCodDh4ArfpAXjZ8cDqs3/6LhGOjg7x4Rhfa1E5h3cZsLntvPNm5+cVcrCRJijSDjqTSIRiEvv+FGu1g42oY2A+yMrY7NCE2ihdO70JKfDQTFq3n/q+mF3OxkiQp0gw6kkqP2CToNxCSqsHKKfDZxZC//dWa+lUSefzkjgC8MWYBX0xaVoyFSpKkSPM9OpJKn0W/wpu9Q9vYKjUKvVg0uRok1wiFoOTqoXb9bjw0YjkvjJhLUmwUgy/fn6bVkyNdvSRJ2gO7mg0MOpJKp4kDYfBlULD9d+sAUK8buf2HcuarY/l53lqa10jms8u6kxgbXXx1SpKkImXQkVT2pS+DtXMhY2XouZ2MVaHPxlUw57tQCLp8HKvj6nP006NYtSGL4zrW5olTOhIIBCJdvSRJ2g27mg3815qSSq+U2qHP9rx7Esz+FqZ8TLVDBvDsaZ3p9/IvfDZxGV0aVubMbg2Kt1ZJklSsPIxAUtnU9sTQn1M/hoIC9m5UmZuPbAnAPV9MY9Li9ZGrTZIkhZ1BR1LZ1PIoiI6HtXNg+UQAzj+gET3b1CA7L5/LB45nc85Onu+RJEmlmkFHUtkUVwFa9Aq1p3wMQCAQ4JGTOlArNZ7F6zJ5bfT8CBYoSZLCyaAjqezasn3t0y3v20mJj+HGI1sA8PwPc1m9IStS1UmSpDAy6Egqu5odDnGpsGEZLBqz5etjO9Shfd1UMrJyeWL4rAgWKEmSwsWgI6nsio6D1n1C7Skfbfk6GAxw69GtAXh/7CJmrEiPRHWSJCmMDDqSyrZ2J4X+nDYYcrO3fL13o8r0aluT/AK4b8h0SsErxSRJUiEYdCSVbQ0PgOQakPknzP1+q66be7UkNirIqNlrGDFrdYQKlCRJ4WDQkVS2BaOgzfGh9tSPt+pqUCWJ/t0bAqFVndy8/GIuTpIkhYtBR1LZ1+6v09dmDIHsjVt1XXZIUyonxTJnVQYDxy6KQHGSJCkcDDqSyr46XaBSQ8jZBDOHbtWVmhDDNT2aAfDE8NmkZeZEoEBJklTUDDqSyr5A4H/v1Jny8Tbd/fauT9PqyazbmM3zP8wp5uIkSVI4GHQklQ9/b1+bMxw2rduqKzoqyC1HtQLg9dELWLR2U3FXJ0mSiphBR1L5UL0V1GgL+Tkw/fNtug9uUY0DmlUlOy+fB4ZOj0CBkiSpKBl0JJUfbU8I/bmd7WuBQIBbjm5FMABDp65gyOTlxVycJEkqSgYdSeXH30FnwU+Qvmyb7pY1U7j4oCYA3PzJZLewSZJUihl0JJUflRpAvX2AAvhj0HaHXHt4c7o0qMSGrFwuHzie7FzfrSNJUmlk0JFUvvx9+trEgZCft013dFSQp/t1IjUhhslL0nj46xnFXKAkSSoKBh1J5UubvhCTCCunwPA7tzukTsUEHj6xPQCv/DSf72esLMYCJUlSUTDoSCpfkqvBsc+G2mOe3u7BBAA929Sk/34NAbjuw0ksT8sspgIlSVJRMOhIKn/angD7XxNqD74clk/a7rABR7WkbZ0U/tyUw1XvTyQ3z+d1JEkqLQw6ksqnQ2+DpodDbia8fzpkrN5mSFx0FM/060xSbBRj56/j6e/nRKBQSZK0Oww6ksqnYBSc8ApUbgJpi+GjsyEvZ5thjaomcf/x7QB45vvZjJm7prgrlSRJu8GgI6n8SqgI/QZCbAVYOBq+HrDdYcd2rMPJXetSUABXvz+RNRlZxVunJEkqNIOOpPKtWgs4/qVQ+7eXYfxb2x125zFtaFY9mVUbsrj+o0nk5xcUY5GSJKmwDDqS1PIoOOSWUPvLa2Hx2G2GJMZG88xpnYiLDjJi5mpe/Wl+MRcpSZIKw6AjSQAHXA8te0N+DnxwBqxfvM2QljVTuK13awAe/mYGkxavL+YiJUnSrjLoSBJAMAh9X4TqrSFjJbx3CmxO32bY6fvUp1fbmuTkFXDFwAls2LztAQaSJCnyDDqS9Le4CnDaB5BUHVb9AR+fA3m5Ww0JBAI8eHx76lRMYNG6TdwyaCoFBT6vI0lSSWPQkaR/qlgfTnsfohNgznAYeiP8vyCTmhjD0/06ERUM8PmkZXz0+5IIFStJknbEoCNJ/1+dLnDCy0AAxr0Kvzy/zZAuDSpx3RHNAbhj8B/MWbWhmIuUJEk7Y9CRpO1p1QeOuCfU/uYWmDFkmyEXH9iE/ZtWJTMnj8vfm8DmnLxiLlKSJO2IQUeSdmTfy6HruUABfHI+LJuwVXcwGODxUzpQNTmWGSs2cN+Q6ZGpU5IkbcOgI0k7EghAr0egyWGQsyl0Etv/O3a6eoV4Hju5IwBv/7KQ4dNWRqBQSZL0/xl0JGlnoqLhpDegepv/HTudk7nVkIOaV+OCAxoBMGDQFP7cmB2BQiVJ0j8ZdCTp38SnbH3s9Jhnthly3REtaFItidUbsrjziz8iUKQkSfong44k7YqK9aDXg6H2T09A2tKtuuNjonjs5I4EAzB44jK+nro8AkVKkqS/GXQkaVe1OR7q7xt6Xmf4ndt0d6xXkUsObgLALYOmsjYjq5gLlCRJfzPoSNKuCgTgyAeAAEz5EBaP3WbIlYc1o2XNCqzdmM3tg93CJklSpBh0JKkwaneCTqeH2kNvgvz8rbrjoqN49KQORAcDDJmynC8mLYtAkZIkyaAjSYV16O0QWwGWjYfJH2zT3bZOKpcd0hSA2wZPZdWGzcVdoSRJ5Z5BR5IKq0INOPD6UHv4nZCVsc2Qyw9tSutaKazflMMtg6ZSUFBQvDVKklTOGXQkaXd0uwQqNYKMFfDT49t0x0QFefyUDsREBRg2bSWfTVy6nZtIkqRwMehI0u6IjoOe94XaY56FPxdsM6RlzRSu7tEcgDsG/8HKdLewSZJUXAw6krS7WhwFjQ6CvCz49rbtDrnowMZ0qJtK+uZcHhw6o5gLlCSp/DLoSNLu+vu46UAQpn8O80dtMyQ6Ksi9x7UjEIBBE5YyYdGfEShUkqTyx6AjSXuiRhvoem6o/fUAyM/bZki7uqmc2LkuAHd/Oc2DCSRJKgYGHUnaUwf/B+JTYeWU7R5MAHBDzxYkxkYxYdF6PvfdOpIkhZ1BR5L2VFIV6PlAqP3D/dvdwlY9JZ5LD24CwENDZ5CZve3KjyRJKjoGHUkqCp1Ohw6nQUE+fHIeZKzaZsj5BzSmTsUElqVt5pVR8yJQpCRJ5YdBR5KKytGPQrWWkLESPjl/m+d14mOiuLlXSwCeHzHX46YlSQojg44kFZXYJDjpTYhJhPkj4cdHthnSu30tujSoRGZOHg9/PTMCRUqSVD4YdCSpKFVvCUf/dSDBiAdh3situgOBALf3bg3AJ+OXMHnJ+mIuUJKk8sGgI0lFrWM/6HQGUBDawrZh5VbdHepV5PhOdQC4+wuPm5YkKRwMOpIUDr0egeqtYeOq0OEE/+95nRuObEFCTBTjFv7JkCnLI1SkJElll0FHksIhNvGv53WSYMGo0Da2f6iVmsDFB4WOm37gqxlszvG4aUmSipJBR5LCpVpz6PNkqP3jI9u8X+fCAxtTKzWepesz+e9Ij5uWJKkoGXQkKZzanwydzgQK4MtrIDd7S1dCbBQDjmoFwHM/zGH2yg0RKlKSpLLHoCNJ4XbEvZBUDdbOhp+f2aqrT/taHNayOtl5+dz0yWTy8j2YQJKkomDQkaRwS6gIh98Tao98BNYv2tIVCAS4t29bkuOiGb9oPW/9vCAiJUqSVNYYdCSpOHQ4FervB7mZ8PWArbpqpSYw4KiWADz89UwWr9sUiQolSSpTDDqSVBwCATj6MQhEwYwvYda3W3X326s++zSqTGZOHv8ZNMV360iStIcMOpJUXGq0hm6XhNpDb4CczC1dwWCAB09oT1x0kFGz1/Dx70siVKQkSWWDQUeSitPBN0OFWvDnAvjpya26GlVN4trDmwNwz5fTWLVhc/HXJ0lSGWHQkaTiFFcBet4fav/0BKydu1X3efs3ol2dVNI353LH4D8iUKAkSWWDQUeSilubvtD4EMjLgqE3wj+ex4mOCvLQCe2JDgYYOnUFQ6csj2ChkiSVXgYdSSpugQAc9ShExcKc4TD9i626W9dO4eKDmgBw2+A/SNuUE4kqJUkq1Qw6khQJVZvCfleG2l8PgOyNW3VffmhTmlRLYk1GFrcNnuopbJIkFZJBR5Ii5YDrILU+pC+B4Xdt1RUfE8XDJ3YgKhjg80nLGDh2cYSKlCSpdDLoSFKkxCZC7ydC7bH/hRlfbdXdpUElbujZAoA7v/iDP5alFXeFkiSVWgYdSYqkZj1g38tD7cGXQtrW78+58IDGHNqyOtm5+Vz27ng2bPZ5HUmSdoVBR5Ii7bA7oHYnyPwTPjkf8nK3dAWDAR47qQO1U+NZsHYTN386xed1JEnaBQYdSYq06Fg48TWIrQCLfoaRD27VXSkplmdP70x0MMCQyct555eFESpUkqTSw6AjSSVB5cZwzFOh9o+PwryRW3V3rl+Jm3u1BOCeL6czZYnP60iStDMGHUkqKdqeAJ3PAgrg0wsgY/VW3eft34jDW9cgOy+fy94bT7rP60iStEMGHUkqSY58CKq1hIyV8NnFkJ+/pSsQCPDoiR2oWymBRes2ceNHk31eR5KkHTDoSFJJEpsIJ74O0fEwZzj8/OxW3amJMTx3WmdiogJ8/ccK3hizIDJ1SpJUwhl0JKmkqdEaej0Uan93Fyz+bavuDvUq8p+jWgHwwFczmL48vbgrlCSpxDPoSFJJ1PlsaNMX8nNh4Kmwdu5W3f33a8hhLauTnZfP1e9PZHNOXoQKlSSpZDLoSFJJFAhAn6ehVgfYtAbeOg7Sl/2jO8BDJ7ananIcM1du4MGhMyJXqyRJJZBBR5JKqvgUOP0TqNwE0hbB28fDpnVbuqsmx/HISe0BeGPMAkbMXBWpSiVJKnEMOpJUkiVXgzMHQYVasHo6vHcKZG/c0n1Ii+r0368hANd/NJm1GVkRKlSSpJLFoCNJJV2lBqGwE18RloyFD8+C3Owt3Tf3aknzGsmsycjipk88clqSJDDoSFLpUL0VnP4RxCSGjp3+7JIt79iJj4niyVM6ERsVZPj0Vbw3dlGEi5UkKfIMOpJUWtTbG05+G4LRMPVj+Pom+Gv1pnXtFG48sgUA93w5jTmrMiJZqSRJEWfQkaTSpFkP6PtfIABjX4LRT27pOrd7I/ZvWpXNOflc/cEEsnPzI1amJEmRZtCRpNKm3Yn/e6HoD/fDuvkABIMBHju5AxUTY5i6NJ3Hhs2MYJGSJEWWQUeSSqO9L4TGB0NeNgy7bcvXNVLiefD40JHTL/04j7Hz1+3gBpIklW0GHUkqjQIB6PkABKJg+hcw/8ctXUe2rclJXepSUADXfTSRjKzcCBYqSVJkGHQkqbSq0Rq6nhtqfz0A8v4XaG7v05o6FRNYvC6T+4ZMi1CBkiRFjkFHkkqzQ/4Ter/Oyqkw/s0tX1eIj+HRkzoAMHDsYr6fsTJCBUqSFBkGHUkqzRIrh8IOwPf3QuafW7r2bVKF8/ZvBMCNH09h3cbs7d1BkqQyyaAjSaVd13OhWkvIXAcjH96q64aeLWhaPZk1GVnc+tkUCv56744kSWWdQUeSSruoGOh5f6g99iVYPWtLV3xMFE+c3JHoYICvpqzg80nLIlSkJEnFy6AjSWVB08OgeS/Iz4Vv/rNVV7u6qVxxaDMAbvtsKsvTMiNRoSRJxcqgI0llRc/7IBgDc4bBrG+36rr0kCZ0qJtK+uZcbvx4slvYJEllnkFHksqKKk1gn4tC7W/+A3k5W7piooI8fkpH4qKDjJq9hnd+WRihIiVJKh4GHUkqSw66ERKrwtrZoed1/qFJtWQG9GoJwH1fTWfBmo2RqFCSpGJh0JGksiQ+FQ67LdQefhfMH7VV91n7NmS/JlXYnJPPDR9PIj/fLWySpLLJoCNJZU2nM6HF0ZCXBQP7wdLxW7qCwQAPndCepNgoflvwJ6+PWRC5OiVJCiODjiSVNcEoOPE1aHgAZG+Ad06A1TO3dNernMh/jm4FwMNfz2De6oxIVSpJUtgYdCSpLIqJh34DoXbn0ItE3zoO1i/a0n3a3vXZv2lVsnLzueHjyeS5hU2SVMYYdCSprIqrAKd/DFVbwIZlobCTsQqAQCDAQye2Jzkumt8X/slrP82PbK2SJBUxg44klWVJVeDMQZBaH9bNhbePh8z1ANSpmMBtvUNb2B75diZzVrmFTZJUdhh0JKmsS60DZ30GSdVh5RR47xTI3gTAyV3rcVDzamTn5nPdR5PIzcuPbK2SJBURg44klQdVmsCZn0JcKiz+BT4+BwoKCAQCPHhCOyrERzNp8XpeHuUWNklS2WDQkaTyomY7OP1DiI6HWV/DxPcAqJWawO29WwPwxLBZzF65IZJVSpJUJAw6klSe1O8Gh/wn1P72Vti4FoATu9Tl0JbVyc4LbWHLcQubJKmU262g89xzz9GwYUPi4+PZZ599GDt27E7HP/nkk7Ro0YKEhATq1avHNddcw+bNm3erYEnSHup2KVRvEzp2ethtQOgUtgeOb0dKfDSTl6Tx1PDZES5SkqQ9U+ig88EHH3Dttddyxx13MH78eDp06EDPnj1ZtWrVdse/99573Hzzzdxxxx1Mnz6dV199lQ8++ID//Oc/e1y8JGk3RMVAnyeBAEx8F+aPAqBGSjz3H98OgOdGzOGXeWsjV6MkSXuo0EHn8ccf54ILLuCcc86hdevWvPjiiyQmJvLaa69td/yYMWPo3r07p512Gg0bNuSII46gX79+/7oKJEkKo3p7Q9dzQ+0vr4HcLAB6t6/NyV3rUlAA13wwkbRNOREsUpKk3VeooJOdnc3vv/9Ojx49/neDYJAePXrw888/b/ea/fbbj99//31LsJk3bx5fffUVRx111A7nycrKIj09fauPJKmIHXY7JNeAtbPhpye3fH1HnzY0qprE8rTNDBg0mYKCgsjVKEnSbipU0FmzZg15eXnUqFFjq+9r1KjBihUrtnvNaaedxt13383+++9PTEwMTZo04eCDD97p1rUHHniA1NTULZ969eoVpkxJ0q5IqAhHPhBqj3oM1swBICkumqdP7URMVICvpqzgw3GLI1ejJEm7Keynro0YMYL777+f559/nvHjx/Ppp58yZMgQ7rnnnh1eM2DAANLS0rZ8Fi/2/8lKUli0OR6a9oC8LBhyDfy1etOubirXHdECgDs/n8bc1RmRrFKSpEIrVNCpWrUqUVFRrFy5cqvvV65cSc2aNbd7zW233caZZ57J+eefT7t27ejbty/3338/DzzwAPn52z++NC4ujpSUlK0+kqQwCATgqEdD79aZ/yNM/mBL14UHNGa/JlXIzMnjyoETyMrNi2ChkiQVTqGCTmxsLF26dOG7777b8l1+fj7fffcd++6773av2bRpE8Hg1tNERUUBuO9bkkqCyo3goJtC7W/+A5vWARAMBnj85I5USozhj2XpPPbtrAgWKUlS4RR669q1117Lyy+/zJtvvsn06dO55JJL2LhxI+eccw4AZ511FgMGDNgyvk+fPrzwwgu8//77zJ8/n2HDhnHbbbfRp0+fLYFHkhRh+10B1VvDprUw7PYtX9dMjeehE9oD8NKP8xg1e3WkKpQkqVCiC3vBKaecwurVq7n99ttZsWIFHTt25Ouvv95yQMGiRYu2WsG59dZbCQQC3HrrrSxdupRq1arRp08f7rvvvqL7LSRJeyYqBno/Ca8dARPehtbHQbPQCZtHtKnJGd3q884vi7j2w0kMveoAqibHRbRcSZL+TaCgFOwfS09PJzU1lbS0NJ/XkaRwGnoT/PoiJFWHS8ZAcjUAMrPzOObZn5i9KoMDmlXlzXP2JhgMRLhYSVJ5tKvZIOynrkmSSpEed4W2sG1cBYMv3XIKW0JsFM+e1pn4mCCjZq/hhZFzI1yoJEk7Z9CRJP1PTDyc8CpExcHsb2Hsy1u6WtSswN3HtAXg8WGzGDt/XaSqlCTpXxl0JElbq9Eajrg31P72Vlg5bUvXSV3r0rdTHfLyC7hy4ATWbcyOUJGSJO2cQUeStK29L4BmPUMvEv3kPMjJBCAQCHDvcW1pXC2JFembufbDieTnl/hHPSVJ5ZBBR5K0rUAAjn0udCjBqmkw7I4tXUlx0Tx3WmfiooOMmLmal0bNi2ChkiRtn0FHkrR9ydXguBdC7bH/hVnfbulqVSuFO49pA8Aj38zk94U+ryNJKlkMOpKkHWvWA/a5JNQefClkrNrSdepe9TimQ23y8gu44r0J/OnzOpKkEsSgI0nauR53Qo22sHE1fHYJ5OcDoed17j++HY2qJrEsbTM3fDyJUvBqNklSOWHQkSTt3N9HTkfHw5zh8NV1W96vkxwXzbOndSI2Osjw6at45vs5ES5WkqQQg44k6d9Vbxk6nIAAjHsNht60Jey0qZ3KPceGntd5fNgsvpy8LIKFSpIUYtCRJO2adif+L+yM/W/oHTt/hZ1T9qrPefs3AuC6DycxafH6yNUpSRIGHUlSYXQ6Hfo8GWr//Cx8d9eWsPOfo1pxaMvqZOXmc/5b41i2PjNydUqSyj2DjiSpcLr0h6MeDbV/egJGPABAVDDAU6d2pEWNCqzekMX5b45jY1Zu5OqUJJVrBh1JUuHtfQEc+WCoPfIhGPkIABXiY3jl7K5USYpl2vJ0rvlgIvn5nsQmSSp+Bh1J0u7pdgkcfk+o/cO9odUdoF7lRF46qwuxUUG+nbaSh7+ZGcEiJUnllUFHkrT7ul8Jh94Wag+/Eya+B0CXBpV5+MT2ALw4ci4fjVscoQIlSeWVQUeStGcOvB4OuC7U/vIaWD4JgOM61eGKQ5sC8J9BU/h13tpIVShJKocMOpKkPXfIrdDsCMjdDB+cCZvWAXBNj+Yc1a4mOXkFXPzO7yxcuzHChUqSyguDjiRpzwWDcPxLUKkhrF8In14A+fkEgwEeO6kj7eum8uemHM594zfSMnMiXa0kqRww6EiSikZCJTjlHYiOhznDYWToVLaE2CheOasrtVLjmbt6I5e/N57cvPwIFytJKusMOpKkolOzHfR5KtQe+RDM/BqA6inxvHJ2VxJjoxg1ew13fvEHBQUeOy1JCh+DjiSpaHU4Ffa6INQedCGsmwdAm9qpPHlKRwIBeOeXRbw5ZkHkapQklXkGHUlS0et5P9TdGzanhQ4nyN4EwBFtanLzkS0BuPvLafwwc1Ukq5QklWEGHUlS0YuOhZPfhKRqsHIqfHk1/LVV7cIDG3Ny17rkF8AV701g5ooNka1VklQmGXQkSeGRUhtOfB0CUTD5Axj7EgCBQIB7j2vHPo0qk5GVy3lv/saajKwIFytJKmsMOpKk8Gl0ABx+d6j99QCYPwqA2OggL57RhYZVElnyZyaXvjOeHE9ikyQVIYOOJCm89r0M2p0EBXnw0dmwfjEAlZJiebX/XlSIi2bsgnXcN2R6hAuVJJUlBh1JUngFAtDnaajZHjathfdP23I4QZNqyTx+SkcA3hizgE/HL4lgoZKkssSgI0kKv9hEOPVdSKwCKybDF1dtOZzg8NY1uPKwZgAM+HQKU5emRbJSSVIZYdCRJBWPivXhpDdDhxNM+RB+fm5L19WHNePQltXJys3nord/Z93G7AgWKkkqCww6kqTi0+gAOPKBUHvYbTD3BwCCwQBPnNKRhlUSWbo+kysGjifXwwkkSXvAoCNJKl57XwgdT4eCfPj4HFg3H4DUhBj+e2ZXEmOjGD1nLY98MzPChUqSSjODjiSpeAUCcPTjULszZP4J758O2RsBaFGzAo+c2AGA//44jy8nL4tkpZKkUsygI0kqfjHxcMo7kFQdVv0B75wIm9YBcHT7Wlx0UGMAbvhoMjNWpEeyUklSKWXQkSRFRmqd0ElscSmwaAy8esSWbWw3HNGC/ZtWJTMnj7NeHcv8NRsjXKwkqbQx6EiSIqfe3nDu15BSF9bOhlcPhyW/Ex0V5Jl+nWhRowKrNmTR76VfWGDYkSQVgkFHkhRZNdrA+cOhZjvYuBreOBqmf0mlpFjevWAfmlVPZkX6Zvq9/AuL1m6KdLWSpFLCoCNJiryUWnDOUGh6OORmwgdnwC8vUDU5jvcu6EbT6sksTwuFncXrDDuSpH9n0JEklQxxFaDf+9DlHKAAvr4Zht5MtaRo3rtgHxpXS2Lp+kxOfekXlvxp2JEk7ZxBR5JUckRFQ+8noMddoZ9/fQEGX0b15DgGXtCNRlVDYaffy7+wdH1mZGuVJJVoBh1JUskSCMD+V8OJr0EgCiYNhB8fpUZKPAMv6EbDKoksXpdJv5d+YXmaYUeStH0GHUlSydT2BDj60VD7h3th6qfUTI1n4IXdqF85kUXrNnHmq2PZsDknsnVKkkokg44kqeTqei50uyzUHnQxLP6NWqkJDLywGzVT4pmzKoNrPphIfn5BZOuUJJU4Bh1JUsl2xD3QvBfkZcH7/eDPhdSpmMB/z+xCbHSQ4dNX8cTwWZGuUpJUwhh0JEklWzAKTnjlf+/ZGXgqbE6nQ72KPNC3HQDPfD+HIZOXR7hQSVJJYtCRJJV8ccnQ7wNIrgmrpsHH50BeLid0qcv5+zcC4PqPJjFtWXqEC5UklRQGHUlS6ZBaB057H6ITYM5w+GYAADf3askBzaqSmZPHhW+PY93G7AgXKkkqCQw6kqTSo3YnOOFlIABjX4JfXyI6Ksgz/TrRoEoiS/7M5LJ3x5OTlx/pSiVJEWbQkSSVLq36QI87Q+1v/gMrplAxMZaXz+pKUmwUP89by31Dpke0RElS5Bl0JEmlT/eroGVvyM+BTy6AnM00r1GBJ07pCMAbYxbwwW+LIlujJCmiDDqSpNInEIA+T0FSdVg9Hb67C4Aj2tTkmh7NAbhl0FS+m74yklVKkiLIoCNJKp2SqsJxz4favzwPc78H4IpDm3Jcx9rk5hdwybvjGTN3TQSLlCRFikFHklR6NTsc9jo/1P7sUti0jmAwwCMndeDw1jXIzs3ngjfHMWHRn5GtU5JU7Aw6kqTS7fB7oEoz2LAcvrgKCgqI+esktu5Nq7AxO4/+r//GjBW+Y0eSyhODjiSpdItNDB05HYyG6Z/DpIEAxMdE8dKZXelcvyJpmTmc8cpYFqzZGOFiJUnFxaAjSSr9aneCg0MvEOWrG+HPBQAkxUXzev+9aVUrhTUZWZz+yq8sW58ZuTolScXGoCNJKhv2vwbqdYPsDfDpRZCfB0BqYgxvnbs3jasmsXR9Jme8+itrMrIiXKwkKdwMOpKksiEYBcf/F2IrwOJfYNRjW7qqVYjj7fP3oXZqPPNWb+Ts18ayMSs3gsVKksLNoCNJKjsqNYSjHg61f7gPRj+9patOxQTevaAbVZNj+WNZOtd8MJH8/ILI1ClJCjuDjiSpbOnQD/a7MtQedhsMuwMKQoGmUdUkXjqrK7HRQb6dtpLHh82KYKGSpHAy6EiSypZAAI64B3rcFfp59JPwxZVbntnpXL8SDx7fDoBnf5jD4IlLI1SoJCmcDDqSpLJp/6uhz9MQCML4t+CjsyE3dAjB8Z3rcvFBTQC44ePJTFy8PnJ1SpLCwqAjSSq7upwNJ70BUbEw/Qt49yTI2gDAjT1b0KNVdbJz87ngrXEsT/PYaUkqSww6kqSyrfWxcPpHEJsM80fCm8fAxrUEgwGePLUTLWpUYPWGLC5863cys/MiXa0kqYgYdCRJZV/jg+HszyGhMiwbD28cDZvTSI6L5pWzu1I5KZYpS9O4/uNJFBR4EpsklQUGHUlS+VCnC5z7NVSoBaunw6BLID+fepUTeeH0zsREBRgyeTlPfzcn0pVKkoqAQUeSVH5UawGnvht6ZmfmEPjpcQD2aVyFe49rC8ATw2fx2QRPYpOk0s6gI0kqX+p0gaMeDbW/vxfmDAfglL3qc8EBjQC44eNJjJm7JlIVSpKKgEFHklT+dDkbOp8NFMAn58OfCwAY0KsVR7erRU5eARe9/TszV2yIaJmSpN1n0JEklU9HPRJa3cn8Ez44A7I3EQwGeOzkDuzVsBIbNudyzutjWZG2OdKVSpJ2g0FHklQ+RcfByW9BYlVYMQWGXAsFBcTHRPHyWV1pUi2JZWmb6f/6WDZszol0tZKkQjLoSJLKr9S6cNLrEAjCpIHw2ysAVEyM5Y1z9qZqchwzVmzg0nfHk5OXH+FiJUmFYdCRJJVvjQ6Ew+8Otb++GRb9CkC9yom83n8vEmOjGDV7DTd/MsV37EhSKWLQkSRp38uhTV/Iz4UPz9pyOEG7uqk8d1pnooIBPhm/hCeGz45snZKkXWbQkSQpEIBjnoXqrSFjBbzZB9YvBuCQltW3vGPn6e9m8+z3s13ZkaRSwKAjSRJAXDKcOQgqN4H1i0JhJ30ZAP32rs81PZoD8Oi3s7h3yHTy8w07klSSGXQkSfpbhZpw9hdQqSH8OT8UdjasBOCqHs249ehWALz603xu+HgyuR5QIEkllkFHkqR/Sq0TCjup9WDtHHjrGNi4BoDzD2jMYyd12PLMzsXvjGdzTl6EC5YkbY9BR5Kk/69ifTj7c6hQG1bPgLeOhU3rADihS11ePKMLsdFBhk9fydmvjSXd9+xIUolj0JEkaXsqNw6t7CTXgJVTQ2En808ADm9dg7fO3ZsKcdH8On8d/V76hTUZWREuWJL0TwYdSZJ2pGpTOOtzSKwKKybD28dvWdnp1rgKAy/sRtXkWP5Yls5JL/7M0vWZES5YkvQ3g44kSTtTvWVoG1tCZVg2Hl7pAWvnAtC2TiofXbwfdSomMH/NRk59ybAjSSWFQUeSpH9Tow30HxI6oGDd3FDYWfgzAI2qJvHxJfvSoEoii9dlGnYkqYQw6EiStCtqtIbzv4PanSFzXeg0tskfAlArNYH3L+xm2JGkEsSgI0nSrqpQI7Sy0+oYyMuGTy+AHx6AggLDjiSVMAYdSZIKIzYRTnoTul8V+nnkgzDoIsjNMuxIUgli0JEkqbCCQTj8bujzFASiYPIHW46fNuxIUslg0JEkaXd16Q9nfAJxqbDoZ3jnBNicbtiRpBLAoCNJ0p5ocgicOzR0/PTS3+G9UyB743bDzjLDjiQVG4OOJEl7qkYbOHPQXys7Y+D90yBn8zZhp9/Lv7A8zbAjScXBoCNJUlGo3RHO+BhikmDeCPjwLMjNplZqAgMv6Ea9ygksXLuJ017+lZXpmyNdrSSVeQYdSZKKSr294bQPIDoeZn8Dn54PebnUrhgKO3UrJTB/zUb6vfQLqww7khRWBh1JkopSowPg1HchKhamDYbBl0J+PnUrJTLwgm7UqZjAvDUbOfXlX1i1wbAjSeFi0JEkqag17QEnvfG/o6e/vBoKCqhXORR2aqfGM2/1Rk57+VdWb8iKdLWSVCYZdCRJCoeWR8MJL0MgCOPfhM8vh7wc6ldJZOCF3aiVGs+cVRmc/sovrMkw7EhSUTPoSJIULm1PgGOfC4WdCe/AwFMhK4MGVZIYeEE3aqTEMWtlBie/+DOL122KdLWSVKYYdCRJCqeOp8Gp70F0AswZDm8cBRtW0rBq0lbP7Bz/whimLk2LdLWSVGYYdCRJCrcWvaD/EEisCssnwas9YPUsGldL5tNL96NlzQqs3pDFqS/9wug5ayJdrSSVCQYdSZKKQ90ucN63ULkxrF8Erx4OC3+mRko8H1y0L90aVyYjK5f+r49l8MSlka5Wkko9g44kScWlShM4bxjU6Qqb18Nbx8Ifn5GaEMOb5+7N0e1rkZNXwFXvT+SVUfMiXa0klWoGHUmSilNSVTj7C2hxNORlwUf94ZcXiIuO4plTO3FO94YA3DtkOvcNmUZ+fkFEy5Wk0sqgI0lScYtNhFPehr3OBwrg65vh29sIUsDtvVszoFdLAF4eNZ/L3htPRlZuZOuVpFLIoCNJUiQEo+CoR+Gw20M/j3kaBl1EIC+Hiw5qwuMndyAmKsDQqSs45tmfmLVyQ2TrlaRSxqAjSVKkBAJwwHVw3AsQiIIpH8J7J0PWBo7vXJcPLtqXWqnxzFu9kWOfHe0hBZJUCAYdSZIireNpcNqHEJME836AN46GDSvpXL8SX16xP/s3rUpmTh5XvT+ROwZPJTs3P9IVS1KJZ9CRJKkkaNYD+n/xj3ftHA5r51IlOY43z92bKw5tCsCbPy/k5P/+zLL1mREuWJJKNoOOJEklRZ2/3rVTqSGsXxgKO4vHEhUMcN0RLXitf1dSE2KYuHg9vZ/5iZ9m+3JRSdoRg44kSSXJ3+/aqdURNq2F13vBmGcgP59DW9bgyyv2p22dFNZtzObs18cycOyiSFcsSSWSQUeSpJImuTr0HwKtj4X8XPj2Vhh4KmxcS73KiXx88X4c36kOefkFDPh0Cg8Mne77diTp/zHoSJJUEsUlw0lvwtGPQ1QczP4GXtwfFowmPiaKx07uwDU9mgPw35HzuHzgeDbn5EW4aEkqOQw6kiSVVIEA7HUeXPAdVGkGG5bBm71hxEMECvK5qkcznjilA7FRQb6asoJ+L//CmoysSFctSSWCQUeSpJKuZju4cAR0OA0K8mHE/fDWsZC+nL6d6vL2eXuTmhDDhEXr6fv8aOas8uWikmTQkSSpNIhLhr4vQN//ht63s2AUvHworJ7FPo2r8Oml+9GgSiKL12Vy/PNjGDPXE9kklW8GHUmSSpMOp8JFI6Fqi9BWttd7wfLJNKmWzKBLu9OlQSXSN+fS/7XfGDJ5eaSrlaSIMehIklTaVG0G5wyFWh1g05rQczuLx1I5KZZ3z9+Ho9vVIjsvn8sHjuftnxdEulpJigiDjiRJpVFSFTj7C6jXDTanwVvHwbwRxMdE8XS/TpzZrQEFBXDb4D94YtgsCgo8flpS+WLQkSSptIpPhTM/hcaHQM5GePdkmDmUqGCAu49tw9U9mgHw1HezufWzqeT5rh1J5YhBR5Kk0iw2CU77AFr2hrwseP90mPIxgUCAq3s0557j2hIIwLu/LuKKgePJyvVdO5LKh90KOs899xwNGzYkPj6effbZh7Fjx+50/Pr167nsssuoVasWcXFxNG/enK+++mq3CpYkSf9PdFzo5aLtT4GCPPjkfPjtVQDO7NaAZ/t13vKunXNe/40Nm3MiXLAkhV+hg84H/9fefUdXVeb7H3+fdGoChBQg9BI6SDNEdIQoNqzXUQe7v3FGUVHujKNjm7FhGedyrSjjqPeOjm2sWBmkiNJEUAQEVDqEThJCCcnZvz+ORnPFewMDnOTwfq111nI/z97J96z1LM3Hvff3eeEFRo0axW233cann35Kz549GTp0KBs2bNjr+WVlZRx33HEsX76cl19+mcWLFzNu3DiaN2/+LxcvSZK+FZ8Ap4+FvpcCAbw1CsaPgvIyTu6RzdOX9KNeUjwff72Zcx6fwZptO6NdsSQdVKFgH99OHDBgAP369ePhhx8GIBwOk5OTw9VXX80NN9zwo/PHjh3L/fffz5dffkliYuJ+FVlcXExqaipFRUU0bNhwv36GJEmHhSCAqffDpLuBAFr0g5//FzRsxhdrirj4qVls2l5G43pJPPKLI8hr1yTaFUvSPqluNtinOzplZWXMmTOHgoKC739AXBwFBQVMnz59r9e88cYb5OXlMWLECDIzM+nWrRt33303FRU//Yzw7t27KS4urvKRJEnVEArBMdfD8JcizQpWz4bHj4bl0+jWPJXXRuTTtVlDtpSWcf6TM/nrtGV2ZJMUk/Yp6GzatImKigoyMzOrjGdmZlJYWLjXa7755htefvllKioqePvtt7nlllt44IEHuPPOO3/y94wePZrU1NTKT05Ozr6UKUmSOhwHl0+GzO5QuhGeORU+fpgWaXX4xxUDOaN3cyrCAbePX8h1L8xjZ5lNCiTFloPedS0cDpORkcETTzxBnz59OOecc7jpppsYO3bsT15z4403UlRUVPlZtWrVwS5TkqTY07gtXPb+900K3r8JXr6UlPBO/vzzntw2rAvxcSFem7eWsx77mFVbdkS7Ykk6YPYp6KSnpxMfH8/69eurjK9fv56srKy9XpOdnU3Hjh2Jj4+vHOvcuTOFhYWUlZXt9Zrk5GQaNmxY5SNJkvZDUl0443E48X6IS4AFr8BfCght/ppL8tvwt8sG0KReEgvXFXPqw9OYtnRTtCuWpANin4JOUlISffr0YeLEiZVj4XCYiRMnkpeXt9dr8vPz+eqrrwiHw5VjS5YsITs7m6SkpP0sW5IkVVsoBAMuh4vfgvqZsHERPPEzWPgGee2a8MbVR9G9eSpbd+zhwr/O5OEPlhJ2c1FJtdw+P7o2atQoxo0bxzPPPMOiRYu44oorKC0t5ZJLLgHgwgsv5MYbb6w8/4orrmDLli2MHDmSJUuW8NZbb3H33XczYsSIA/ctJEnS/63lkfCrqdByIJSVwIsXwPu30LxBIi/9Oo9/69OCcAB/en8Jlz4zm62le3/yQpJqg30OOueccw5/+tOfuPXWW+nVqxfz5s3j3XffrWxQsHLlStatW1d5fk5ODu+99x6zZ8+mR48eXHPNNYwcOXKvraglSdJB1iALLnoD8q6KHH/8IPz36aTs3syfzu7JfWf1IDkhjsmLN3Lygx8yZ8XW6NYrSftpn/fRiQb30ZEk6SBY8Cq8fhWUbYcG2XD2M9ByAIvWFXPls5+ybFMpCXEhbjgxl8uOakMoFIp2xZJ0cPbRkSRJMaTrGfDLSZDeCUrWwdMnwfRH6ZxZjzeuyufkHtmUhwPufGsRv/7bHIp27ol2xZJUbQYdSZIOZ007wi8/iISecDm8dyM8fgwN1kzj4fN6c8dpXUmKj+O9BesZ9tA0ZnyzOdoVS1K1+OiaJEmCIIDZf4GJd8DuoshY+wI47g4+39OMK5/9lNVbdwJwbr8cbjyxM6l1E6NYsKTDVXWzgUFHkiR9b8cWmHIfzB4XucMTioPe51My8HpGf7iN52auBCC9fjJ/OLULJ3fP9t0dSYeUQUeSJO2/zV/DxD/Cwtcjx4l1YeA1fNLyYm54fQlfbdgOwODcDO44vRvN0+pEsVhJhxODjiRJ+tetnAnv3wyrZ0WOW/Rj95lP8dinO3l00teUVYSpmxTPb47vxEUDWxMf590dSQeXQUeSJB0YQQALXoHx18GuIqjXFM5+hq/q9uDGV+Yze3lkr53+bRrz55/3pEWjulEuWFIss720JEk6MEIh6HYWXD4ZMrtB6UZ4Zhjtv/5vXvjlkdx9RnfqJcUza9kWTvzPD3l93ppoVyxJBh1JklRNjdvCZROg+88hqID3biTu1V/yi95NeHvkII5omUbJrnJGPj+Pa/4+1313JEWVQUeSJFVfUl048wk48T6IS4AvXoa/HEcrCnnxV3lcV9CR+LgQb3y2lhPHTHXfHUlRY9CRJEn7JhSCAb+Ci8ZD/UzYsACeOJaErycwsqADL/06j1ZN6rK2aBfnjZvBPe98SVl5ONpVSzrMGHQkSdL+aZUHl0+BnAGRTUafOwcm38sRLVJ5+5pBnNM3hyCAsVO+5twnplNYtCvaFUs6jBh0JEnS/muYHbmz0++XQACT74bnz6NeeDv3/lsPxp5/BA1TEvh05TZOeehDH2WTdMgYdCRJ0r8mIQlO/hOc/hgkpMCSd2HcYFi/kBO6ZfPm1UeRm9WATdvLGP6Xmfzlw2+oBbtbSKrlDDqSJOnA6PULuPQ9SG0JW76GvxTAF6/Qqkk9Xr0yn9N7NaMiHHDnW4u4+u9zKd1dHu2KJcUwg44kSTpwmvWK7LfT9mewpxRevgTev5k68QH/cU4v/jCsCwlxIcZ/vo4zHv2IZZtKo1ywpFhl0JEkSQdWvSYw/B+Qf23k+OOH4MkCQusXcHF+G/5++ZE0bZDMkvXbOfWhaYz/fK2Pskk64Aw6kiTpwItPgOP+CGc/A8mpsHYuPHEMfHAX/VrU462rj6Jf60aU7C7nqufm8uu/zWFDiV3ZJB04Bh1JknTwdD0dRsyE3FMgXA5T74PHjyajaD7P/fJIrhncnoS4EO8tWM9xf57KP+as9u6OpAMiFNSCf5sUFxeTmppKUVERDRs2jHY5kiRpXwUBLHwd3v4NlG4EQnDkFTD4ZhZuquD6f3zGF2uKATimY1PuPrM7zdPqRLdmSTVSdbOBd3QkSdLBFwp9e3dnFvQ8DwhgxqPw6JF02T2P167M5/oTOpGUEMeUJRsZ+h9TeXbmCsLhGv//YyXVUN7RkSRJh97Sf8L4a6FoVeT4yCthyK18tbWC61/+jE9XbosMt23M6DN70Ca9XtRKlVSzVDcbGHQkSVJ07C6B92+BOU9FjpvmwhmPU5HVk2c+Xs797y1m554KkhLiGDmkA5cf3ZbEeB9GkQ53Bh1JklQ7LHkf3rgKtq+HuAT42Q2Qfx2risr4/avz+XDpJgBysxpwz1k96JWTFt16JUWVQUeSJNUepZsjj7IteiNy3KI/nDGWoHFbXpu3htvfXMjWHXuIC8HFA9vw78d3pF5yQlRLlhQdBh1JklS7BAF8/gK8/VvYXQyJdWHwLdDv/7F5V8Cdby3i1blrAGieVoc7z+jGsZ0yoly0pEPNoCNJkmqnbavgtStg+YeR48ZtoeAP0PlUJi/ZyE2vfsGabTsBOKFrFrcO60IzW1FLhw2DjiRJqr3CYfj0aZh097f77gA5A+D4uyjN6M1/TFjCUx8vpyIcUCcxnmuGdOCyo9qQlGCzAinWGXQkSVLtt7sEPnoQPn4IyiN3cehyOhTcxpdl6dzy2hfMXr4VgPYZ9bn9tK4MbJcevXolHXQGHUmSFDuK18Kku2Dus0AAcYnQ/3KCo3/LK4tKGf3OIjZtLwPgtF7NuOmkzmQ0TIluzZIOCoOOJEmKPYVfwIRb4euJkeM6jWHwTRR1Gc4D//ya/56xgiCA+skJXD24PRfntyY5IT66NUs6oAw6kiQpdn31T3jvJtj4ZeQ4owsMvZv5yUdw8+tf8NmqbQC0blKXm0/uwpDOGYRCoejVK+mAMehIkqTYVlEOc56KPNK2M/KeDp1OIlxwB6+uTOGed79kY8luAAZ1SOfWU7rQIbNBFAuWdCAYdCRJ0uFhxxaYci/MGgdBReT9nSN/zfa83/DIR4U8+eEyyirCxMeFuODIVlxb0IG0uknRrlrSfjLoSJKkw8vGxZHH2b6aEDlu2AJOuo8VTX/GXW8t4v2F6wFoVDeR24Z15bRezXycTaqFDDqSJOnwtOQ9ePs3sG1l5LjTSXDifUzbWIfbxy9gyfrtAAzOzeCuM7qRnepmo1JtUt1s4K5akiQptnQcClfOhKNGQVwCLH4bHunPURueY/yVRzLquI4kxof44MsNHP/nqTw3cyXhcI3//76S9pF3dCRJUuzasAjGXwcrp0eOM7vBKWNYkpTL9S9/zrxvu7Md2bYx957Vg1ZN6kWvVknV4qNrkiRJAOEwzHsWJtzyfXe2HudSMfhWnpq/mz+9v5hde8KkJMbx78d14uL81iTG+9CLVFMZdCRJkn6odFNks9F5z0aOE+vCUaNYlXspv3tjKR9/vRmA9hn1ueWULhzTsWkUi5X0Uww6kiRJe7N6Drx7A6yeFTlObUlw3B95obQP9763mK079gCRZgU3ndyZdk3rR7FYSf+TQUeSJOmnBAF88Y/IHZ7iNZGxlgMpOfYOxiyoyzMfL6c8HJAQF+LCvNaMHNKB1LqJ0a1ZEmDQkSRJ+r+V7YCPH4RpY6B8JxCC3uezvNe/c/ukTXzw5QYgsvfOqOM7cV6/HBJ8f0eKKoOOJElSdRWtgX/+Aea/GDlOagDHXM/UxmdxxztfsXRDZO+d9hn1ueGEXIZ0znCzUSlKDDqSJEn7auVMePd3sHZu5LhxOyqOv4u/bc5lzMSlle/vDGjTmN+f1JmeOWnRq1U6TBl0JEmS9kc4DJ/9PXKHpzTy6BrtC9h+7B08Mj+Ov05bxu7yMADDejbj+qGdyGlcN3r1SocZg44kSdK/YlcxfPgAzHgUKsogFA9dz2BD919yz2cpvDp3DUEASfFxXJDXiit+1o70+snRrlqKeQYdSZKkA2Hz1/D+zbD47e/HWh3Fyk4Xc/PCFkz9agsAyQlxnN23BZcPakfLJt7hkQ4Wg44kSdKBtO4zmP5IpC11uDwy1qQ9S9peyE3LujF79S4A4kJwUvdsfn1MO7o1T41iwVJsMuhIkiQdDEVrYNbj8MnTsLsIgKBOY5Z3vYrb1w9k0tItlacO6pDOFce0I69dE7u0SQeIQUeSJOlg2l0Cc/8WeYdn28rIWEYXlvW/jf9Ymsn4z9cS/vavrLy2Tbh1WBc6Z/t3jPSvMuhIkiQdChXlMPe/YOLtsHNrZKzrGaztfxNj5+3m+dmrKCsPExeC8/q3ZNRxHWli0wJpvxl0JEmSDqUdW2DSXfDJXyEIQ2JdGDSKVbmXcc+E5bw1fx0ADVISuLagIxfmtSIxPi7KRUu1j0FHkiQpGtZ9Du9cDyunR44btYFBo5hd71hue3c5C9cVA9C2aT1uOaULx3bKiGKxUu1j0JEkSYqWIID5L8OEW6AkcieH5IaEu/+c91JO4ObpsLm0DIA+rRpxwZGtOLF7FskJ8VEsWqodDDqSJEnRtrsEZj8Jc56Grcsqh8ub9eWd5BP4/ZL2lFQkAdC4XhJn923B8P6t3IdH+l8YdCRJkmqKcBiWTYE5T8GXb1XuwxNOTmV2xtncsH4wyyJPtBEKwTEdm3L+gFYcm5tBfJxtqaUfMuhIkiTVRCXrYd7fInd5vm1LHTRoxvzO13H/2h58+NXmylPbZ9Tn1lO6cHTHplEqVqp5DDqSJEk1WTgMi96IvMfz3T48LfqxJu8PPL28MS9+spqinXsAGJKbwU0nd6Zt0/pRLFiqGQw6kiRJtcGeXTDjEZj6AOwpjYz1/AXF+b9nzMwS/mv6csrDAYnxIS4e2Jqrh3SgYUpidGuWosigI0mSVJsUr4OJf4TP/h45TqwHA6/mm3bDuWNiIZMWbwSgSb0kfju0E2f3zfH9HR2WDDqSJEm10eo58O7vYPXsyHFiPehzMR9nnMPNk7byzcbIXZ/2GfW5JL81Z/RuTt2khCgWLB1aBh1JkqTaKghg4Wvw4Z+h8PPIWFwi4e4/55W6/8bt0/dQvCvSua1hSgLn9MvhwrzW5DS2LbVin0FHkiSptgsC+PoDmPYfsPzDbwdD7Ol4Eu+knssDCxuwYvOOyGgICjpncvHA1gxs14RQyMfaFJsMOpIkSbFk1Wz4aAx8Ob5yKGh1FJ+1uogHvmlZpS11blYDrhrcnpO6ZRPnezyKMQYdSZKkWLRxMUwbA/NfrNx4lMxuFHb/FWM39eDFTwvZUVYBQIeM+lw9pAMnd8+2cYFihkFHkiQplhWthumPRjYe/a4tdWpLdva7gidL83l8eiEl377H0z6jPlcPbs8pPZoZeFTrGXQkSZIOBzu2wCdPwoyxsGNTZKxOY3b1/RVP7zmOR2dsqmxc0LZpPa78WXtO7JZFvWQ7tal2MuhIkiQdTvbshHnPwscPwdblkbGkBuw+4hL+OziZh2YVU7RzDwApiXEMyc3klB7ZHJubQUpifPTqlvaRQUeSJOlwVFH+bWvqB2DDwshYQgplPX7B84ln8OQXFZWd2gDqJcVzXJdMTunRjEEd00lOMPSoZjPoSJIkHc7CYVj6Hkz9E6z5JDIWiifochqr04/i9W3t+PviMGu27ay8pGFKAmf0bs45/VrSpZl/c6lmMuhIkiQpshfP8g8jd3i+mVx1qnFbNqUPYMqeXP6yqgVfbq9TOdejRSrn9Mvh1J7NaJCSeIiLln6aQUeSJElVrfkUFr4eCT5r50IQrjK9PS2XfySfzujV3dhVEQdAncR4Tu6Rzbn9cujTqpEbkSrqDDqSJEn6abuKYMV0WDYVlk+FwvmVUxVpbZiWfTF3rurO0k27Kse7ZDfkkvzWDOvZzAYGihqDjiRJkqqvdDPM/W/4+EHYsRmAoFFrVnS5gke39uP1zzewuzxyB6hJvSSGD2jJ+Ue2IqNhSjSr1mHIoCNJkqR9t3t7ZF+ejx78fl+etFaUDhjJszvyeGrWOtYVRe7yJMaHOLl7Npfkt6FnTlr0atZhxaAjSZKk/VdWCp/8FT76TyjdGBmr24Rwr/OZ3GAYj87bwycrtlae3jMnjQuPbMXJPbJ9rE0HlUFHkiRJ/7qyHTDnKZj+KBSv/nYwBB2HsqzNeTy8oiVvzC9kT0XkT8rG9ZL4ed8chg9oSU7jutGrWzHLoCNJkqQDp6I8si/PrHHwzaTvxxu1prT7hbyyqw9jP/9+X55QCAZ3yuD8vFYc1T6dxPi4KBWuWGPQkSRJ0sGx6avIezxzn4XdRZXDQaM2rG58JP/Y1oG/rsmhmHoAJCfE0a15Kr1y0uiZk0bvnDRaNKpjq2rtF4OOJEmSDq6yUvjiHzDv77B6FoTLK6eCUByr63bh7dJcJu7uzLygPWV8v/Fo43pJ9GyRyvFdszi9V3PqJPlej6rHoCNJkqRDZ1cxLJ8Weazt60mweWmV6Yq4ZFbU7cr0cBfGF7djTnnbyuCTWieRc/vncGFea5qn1YlG9apFDDqSJEmKnm2rIqHnm8mw7EMo3VBlOhyfwtqGPZhUksPsHdksClqyMpTNkK7NuSS/DX1bNfLRNu2VQUeSJEk1QxDApqWwfGrkrs/yad+3rP6B3UECXwfNWRTksK1BR3L6n8bR+YNsV60qDDqSJEmqmYIANi6GFR9B4XxYvwA2LISy7VVOCwchXg/9jJU9r+O0QX1pnV4vOvWqRjHoSJIkqfYIh6FoJaxfyI7Vn7Fx4Ye02vIRADuCZB4vP4UFrS/k5wNzGZybQYLtqg9bBh1JkiTVahUrZrL9jetJ3TwPgMKgEX8q/zkf1zuOs/q2ZHBuBj1bpBEX57s8hxODjiRJkmq/IIAFr1L+/m0kFK8EYEG4FXeUX8CMcBea1EvimE5NGZybwaAOTUmtk/h//EDVdgYdSZIkxY7y3TDzcYKp9xPaXQzAG8HR/GH3L9hC5O/D+LgQfVs14qTu2ZzbP4fkBJsYxCKDjiRJkmJP6WaYfDfMfhII2JOUxnvNRzBmU3++2lhaeVrrJnW5+eQuDOmcYZvqGGPQkSRJUuxaPQfeHAnr50eOW+WzbtDdvF2YytgpX7OxZDcAx3Rsyi2ndKF9Rv0oFqsDyaAjSZKk2FZRDjPHwqS7YM8OiEuEo65j+4CRPDR1FX+dtow9FQEJcSEuHtiaawo60DDFd3hqO4OOJEmSDg/bVsLbv4Ul70aOU1vCgMtZ0fIMbv/nOiZ+uQGA9PpJjDquE6f1aka95IQoFqx/hUFHkiRJh48ggEVvwjvXQ8m6yFhCHehxNp9knMX1H8E3377Dk5IYx5DOmQzr0YyfdWpKSqJNC2oTg44kSZIOP2U7YP5LMGvc9+/vAOEWA5iSdgZ3fdOOr7bsqRxvkJzA8V2zGNYzm/z26SS6EWmNZ9CRJEnS4SsIYOUMmPUELHoDwuWR4XpN2dLyBN4NBjD2m0xWFX8fehrVTWRo1yxO6p5NXrsmhp4ayqAjSZIkAZQUwpyn4ZO/wvb1lcNB3SZsanEcb1f059HlzVhfGq6cS6ubyNAuWZzUI5uBhp4axaAjSZIk/VDFHlg2BRa8Bl+Oh51bK6eClDQ2NBvM1D2d+a81zZi/Iw2I7L+TWieR47tkcmxuBvnt0kmta+e2aDLoSJIkST+lYg8snwYLX480Mdixqcr07rpZLE7qytvFbfhgZweWBs0JiCMuBD1z0hjUoSlHd0inV04aCd7tOaQMOpIkSVJ1hCtgxcew9H1YOR3Wzq18p+c7O+IbMjWuH38r7c/0cBcqiHRqa5CcQF67JpzQLYvjumTSwH16DjqDjiRJkrQ/ynbAmk8i4WfFx7B6dmRD0m/tSmrMrDqDeKq4D5N3tiUgckcnKSGOwZ0yGNazGYNzM6iTZNvqg8GgI0mSJB0IFXsiHdwWvBJ51G3H5sqpsrpZzE8bzBPb+vDeliy+e6+nblI8x3WJ7NUzqGM6yQmGngPFoCNJkiQdaN81NPjilci7PbuLK6d2p7ZjVoMhPLSxF7OK0irH6ycn8LNOTRnaNYtjczOon5wQhcJjh0FHkiRJOpjKd8NXEyMblC5+G8p3VU5tb9qbqSmDGbOuK0u2p1SOJ8XHkd++CUO7ZlHQJZP0+snRqLxWM+hIkiRJh8ruElg0Hua/CN9MhiCyJ08QiqckeyAfJ+UxbkNn5mz5PtiEQtCvdWNO6pbFCd2yyUpN+Ykfrh8y6EiSJEnRULIevvhHJPSsnVs5HBBiV1Yf5tQ9iqe3dOOfhXWrXHZEyzRO6p7NCd2yaNGo7v/8qfqWQUeSJEmKts1fR97lWfRmpJPbD5Sld2Vh/SN5fVs7nitsxm6SKud6tkjl+K5ZHN2hKV2bNSQuLnSoK6+xDDqSJElSTVK8Fr58Cxa9Acs/gqCiciqIT6awYQ+m7OnCi1va8Hm4LeVEmhY0qZfE0R2bckzHphzVIf2wf6/HoCNJkiTVVDu2wJJ3I+/zLJsKJeuqTO+Jr8eC5J68sr0b75T1ZCONKue6N0/l6I7pDOrQlCNaNiIpIe4QFx9dBh1JkiSpNggC2PzV96Fn+Yewc2uVU9bV68w/K47g+aKuLAha8cP9evLaNmFQh3QGdWxK2/R6hEKx/ZibQUeSJEmqjcJhKPwclr4fueuzZk6V6R0pmXyW3Id3i9syaVc7VgYZfBd8mqfVIb99Ewa0acKAto1jsqmBQUeSJEmKBSXrYel7sPhd+GYS7NlRZbo0KZ358Z2ZsL0tM8o7sShoSZjI42zN0+rQv01jBrRpTP82jWkTA3d8DDqSJElSrNmzK/Jo2/IPYcX0SPvq8J4qp+yMb8D0xAE8X9KLKRXdqnRzy2iQzJDOGRR0ziS/fTopifGH+hv8yww6kiRJUqzbsxPWfAorp0c+q2bB7uLK6fKE+nyVNpC3yvvzzKYOFJcnVs7VSYxnUId0CrpkMiQ3gya1pJubQUeSJEk63IQrYOWMSAvrhW9AydrKqSCxLpuzBjE11Je/FLZnYfH3wSYUgiNaNiK/fTpHtm3MES0b1di7PQYdSZIk6XAWDkcaGSx8LRJ8tq2snAoIsTPzCOalDOC5bZ0Zv74x3zU0AEhKiKN3ThpHtm1CXrsm9MpJqzHBx6AjSZIkKSIIYN1nsPhtWPxOpKvbD5Q3aM7ytIF8sqcVEzY1Yfr2THaQUjmfnBDHpUe14Xcn5B7qyn+kutkg4RDWJEmSJCkaQiFo1ivyOfb3ULwWlrwX+XwzmYSSNbQveYn2wLkAKbC9TnO+iW/N7B3ZfLqrGS2oB0Q/6FTXfm2j+sgjj9C6dWtSUlIYMGAAs2bNqtZ1zz//PKFQiNNPP31/fq0kSZKkA6FhM+h7CfziefjdMvjFS5B3FbQbDPWzAKi/cw09tn/EZeGXeSTpQc7c8VKUi943+3xH54UXXmDUqFGMHTuWAQMGMGbMGIYOHcrixYvJyMj4yeuWL1/Ob37zGwYNGvQvFSxJkiTpAEqsAx2Pj3y+s2MLrF8Q+WxYAOsXUqd1/+jVuB/2+R2dAQMG0K9fPx5++GEAwuEwOTk5XH311dxwww17vaaiooKjjz6aSy+9lA8//JBt27bx2muvVft3+o6OJEmSJKh+NtinR9fKysqYM2cOBQUF3/+AuDgKCgqYPn36T153++23k5GRwWWXXVat37N7926Ki4urfCRJkiSpuvYp6GzatImKigoyMzOrjGdmZlJYWLjXa6ZNm8aTTz7JuHHjqv17Ro8eTWpqauUnJydnX8qUJEmSdJjbr2YE1VVSUsIFF1zAuHHjSE9Pr/Z1N954I0VFRZWfVatWHcQqJUmSJMWafWpGkJ6eTnx8POvXr68yvn79erKysn50/tdff83y5csZNmxY5Vg4HI784oQEFi9eTLt27X50XXJyMsnJyT8alyRJkqTq2Kc7OklJSfTp04eJEydWjoXDYSZOnEheXt6Pzs/NzWX+/PnMmzev8nPqqady7LHHMm/ePB9JkyRJknRQ7HN76VGjRnHRRRfRt29f+vfvz5gxYygtLeWSSy4B4MILL6R58+aMHj2alJQUunXrVuX6tLQ0gB+NS5IkSdKBss9B55xzzmHjxo3ceuutFBYW0qtXL959993KBgUrV64kLu6gvvojSZIkSf+rfd5HJxrcR0eSJEkSHKR9dCRJkiSpNjDoSJIkSYo5Bh1JkiRJMcegI0mSJCnmGHQkSZIkxRyDjiRJkqSYY9CRJEmSFHMMOpIkSZJijkFHkiRJUswx6EiSJEmKOQYdSZIkSTHHoCNJkiQp5hh0JEmSJMUcg44kSZKkmGPQkSRJkhRzDDqSJEmSYo5BR5IkSVLMMehIkiRJijkGHUmSJEkxx6AjSZIkKeYYdCRJkiTFHIOOJEmSpJhj0JEkSZIUcww6kiRJkmJOQrQLqI4gCAAoLi6OciWSJEmSoum7TPBdRvgptSLolJSUAJCTkxPlSiRJkiTVBCUlJaSmpv7kfCj4v6JQDRAOh1m7di0NGjQgFApFtZbi4mJycnJYtWoVDRs2jGotqj1cN9pfrh3tD9eN9ofrRvvrUK+dIAgoKSmhWbNmxMX99Js4teKOTlxcHC1atIh2GVU0bNjQfwlon7lutL9cO9ofrhvtD9eN9tehXDv/252c79iMQJIkSVLMMehIkiRJijkGnX2UnJzMbbfdRnJycrRLUS3iutH+cu1of7hutD9cN9pfNXXt1IpmBJIkSZK0L7yjI0mSJCnmGHQkSZIkxRyDjiRJkqSYY9CRJEmSFHMMOvvgkUceoXXr1qSkpDBgwABmzZoV7ZJUw4wePZp+/frRoEEDMjIyOP3001m8eHGVc3bt2sWIESNo0qQJ9evX56yzzmL9+vVRqlg10T333EMoFOLaa6+tHHPdaG/WrFnD+eefT5MmTahTpw7du3fnk08+qZwPgoBbb72V7Oxs6tSpQ0FBAUuXLo1ixYq2iooKbrnlFtq0aUOdOnVo164dd9xxBz/sTeW6EcDUqVMZNmwYzZo1IxQK8dprr1WZr8462bJlC8OHD6dhw4akpaVx2WWXsX379kP2HQw61fTCCy8watQobrvtNj799FN69uzJ0KFD2bBhQ7RLUw0yZcoURowYwYwZM5gwYQJ79uzh+OOPp7S0tPKc6667jjfffJOXXnqJKVOmsHbtWs4888woVq2aZPbs2Tz++OP06NGjyrjrRv/T1q1byc/PJzExkXfeeYeFCxfywAMP0KhRo8pz7rvvPh588EHGjh3LzJkzqVevHkOHDmXXrl1RrFzRdO+99/LYY4/x8MMPs2jRIu69917uu+8+HnroocpzXDcCKC0tpWfPnjzyyCN7na/OOhk+fDgLFixgwoQJjB8/nqlTp3L55Zcfqq8Agaqlf//+wYgRIyqPKyoqgmbNmgWjR4+OYlWq6TZs2BAAwZQpU4IgCIJt27YFiYmJwUsvvVR5zqJFiwIgmD59erTKVA1RUlISdOjQIZgwYUJwzDHHBCNHjgyCwHWjvfvd734XHHXUUT85Hw6Hg6ysrOD++++vHNu2bVuQnJwc/P3vfz8UJaoGOvnkk4NLL720ytiZZ54ZDB8+PAgC1432DgheffXVyuPqrJOFCxcGQDB79uzKc955550gFAoFa9asOSR1e0enGsrKypgzZw4FBQWVY3FxcRQUFDB9+vQoVqaarqioCIDGjRsDMGfOHPbs2VNlLeXm5tKyZUvXkhgxYgQnn3xylfUBrhvt3RtvvEHfvn05++yzycjIoHfv3owbN65yftmyZRQWFlZZN6mpqQwYMMB1cxgbOHAgEydOZMmSJQB89tlnTJs2jRNPPBFw3ah6qrNOpk+fTlpaGn379q08p6CggLi4OGbOnHlI6kw4JL+lltu0aRMVFRVkZmZWGc/MzOTLL7+MUlWq6cLhMNdeey35+fl069YNgMLCQpKSkkhLS6tybmZmJoWFhVGoUjXF888/z6effsrs2bN/NOe60d588803PPbYY4waNYrf//73zJ49m2uuuYakpCQuuuiiyrWxt/92uW4OXzfccAPFxcXk5uYSHx9PRUUFd911F8OHDwdw3ahaqrNOCgsLycjIqDKfkJBA48aND9laMuhIB8mIESP44osvmDZtWrRLUQ23atUqRo4cyYQJE0hJSYl2OaolwuEwffv25e677wagd+/efPHFF4wdO5aLLrooytWppnrxxRd59tlnee655+jatSvz5s3j2muvpVmzZq4bxRwfXauG9PR04uPjf9ThaP369WRlZUWpKtVkV111FePHj2fSpEm0aNGicjwrK4uysjK2bdtW5XzX0uFtzpw5bNiwgSOOOIKEhAQSEhKYMmUKDz74IAkJCWRmZrpu9CPZ2dl06dKlyljnzp1ZuXIlQOXa8L9d+qHf/va33HDDDZx77rl0796dCy64gOuuu47Ro0cDrhtVT3XWSVZW1o+adpWXl7Nly5ZDtpYMOtWQlJREnz59mDhxYuVYOBxm4sSJ5OXlRbEy1TRBEHDVVVfx6quv8sEHH9CmTZsq83369CExMbHKWlq8eDErV650LR3GhgwZwvz585k3b17lp2/fvgwfPrzyn103+p/y8/N/1L5+yZIltGrVCoA2bdqQlZVVZd0UFxczc+ZM181hbMeOHcTFVf3zLz4+nnA4DLhuVD3VWSd5eXls27aNOXPmVJ7zwQcfEA6HGTBgwKEp9JC0PIgBzz//fJCcnBw8/fTTwcKFC4PLL788SEtLCwoLC6NdmmqQK664IkhNTQ0mT54crFu3rvKzY8eOynN+/etfBy1btgw++OCD4JNPPgny8vKCvLy8KFatmuiHXdeCwHWjH5s1a1aQkJAQ3HXXXcHSpUuDZ599Nqhbt27wt7/9rfKce+65J0hLSwtef/314PPPPw9OO+20oE2bNsHOnTujWLmi6aKLLgqaN28ejB8/Pli2bFnwyiuvBOnp6cH1119feY7rRkEQ6QQ6d+7cYO7cuQEQ/PnPfw7mzp0brFixIgiC6q2TE044Iejdu3cwc+bMYNq0aUGHDh2C884775B9B4POPnjooYeCli1bBklJSUH//v2DGTNmRLsk1TDAXj9PPfVU5Tk7d+4MrrzyyqBRo0ZB3bp1gzPOOCNYt25d9IpWjfQ/g47rRnvz5ptvBt26dQuSk5OD3Nzc4IknnqgyHw6Hg1tuuSXIzMwMkpOTgyFDhgSLFy+OUrWqCYqLi4ORI0cGLVu2DFJSUoK2bdsGN910U7B79+7Kc1w3CoIgmDRp0l7/prnooouCIKjeOtm8eXNw3nnnBfXr1w8aNmwYXHLJJUFJSckh+w6hIPjBVriSJEmSFAN8R0eSJElSzDHoSJIkSYo5Bh1JkiRJMcegI0mSJCnmGHQkSZIkxRyDjiRJkqSYY9CRJEmSFHMMOpIkSZJijkFHknTIXXzxxZx++unRLkOSFMMMOpIkSZJijkFHklSjTJkyhf79+5OcnEx2djY33HAD5eXllfMvv/wy3bt3p06dOjRp0oSCggJKS0sBmDx5Mv3796devXqkpaWRn5/PihUrovVVJElRZNCRJNUYa9as4aSTTqJfv3589tlnPPbYYzz55JPceeedAKxbt47zzjuPSy+9lEWLFjF58mTOPPNMgiCgvLyc008/nWOOOYbPP/+c6dOnc/nllxMKhaL8rSRJ0ZAQ7QIkSfrOo48+Sk5ODg8//DChUIjc3FzWrl3L7373O2699VbWrVtHeXk5Z555Jq1atQKge/fuAGzZsoWioiJOOeUU2rVrB0Dnzp2j9l0kSdHlHR1JUo2xaNEi8vLyqtyFyc/PZ/v27axevZqePXsyZMgQunfvztlnn824cePYunUrAI0bN+biiy9m6NChDBs2jP/8z/9k3bp10foqkqQoM+hIkmqN+Ph4JkyYwDvvvEOXLl146KGH6NSpE8uWLQPgqaeeYvr06QwcOJAXXniBjh07MmPGjChXLUmKBoOOJKnG6Ny5M9OnTycIgsqxjz76iAYNGtCiRQsAQqEQ+fn5/PGPf2Tu3LkkJSXx6quvVp7fu3dvbrzxRj7++GO6devGc889d8i/hyQp+nxHR5IUFUVFRcybN6/K2OWXX86YMWO4+uqrueqqq1i8eDG33XYbo0aNIi4ujpkzZzJx4kSOP/54MjIymDlzJhs3bqRz584sW7aMJ554glNPPZVmzZqxePFili5dyoUXXhidLyhJiiqDjiQpKiZPnkzv3r2rjF122WW8/fbb/Pa3v6Vnz540btyYyy67jJtvvhmAhg0bMnXqVMaMGUNxcTGtWrXigQce4MQTT2T9+vV8+eWXPPPMM2zevJns7GxGjBjBr371q2h8PUlSlIWCHz4fIEmSJEkxwHd0JEmSJMUcg44kSZKkmGPQkSRJkhRzDDqSJEmSYo5BR5IkSVLMMehIkiRJijkGHUmSJEkxx6AjSZIkKeYYdCRJkiTFHIOOJEmSpJhj0JEkSZIUcww6kiRJkmLO/wcdhs1vZv+aEgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import  matplotlib.pyplot as plt\n",
    "plt.figure(figsize=(10,10))\n",
    "plt.plot(firtting.history['loss'])\n",
    "plt.plot(firtting.history['val_loss'])\n",
    "plt.legend(['train Loss','test Loss'],loc='upper left')\n",
    "plt.xlabel('Epochs')\n",
    "plt.xlabel('Loss')\n",
    "plt.show()\n",
    "#-----------------------------#\n",
    "plt.figure(figsize=(10,10))\n",
    "plt.plot(firtting.history['accuracy'])\n",
    "plt.plot(firtting.history['val_accuracy'])\n",
    "plt.legend(['train accuracy','test accuracy'],loc='upper left')\n",
    "plt.xlabel('Epochs')\n",
    "plt.xlabel('accuracy')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fusion Using Extra Layer\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of frames: (29, 64, 64, 3)\n",
      "CSV data:\n",
      "(29, 51)\n"
     ]
    }
   ],
   "source": [
    "import cv2\n",
    "import pandas as pd\n",
    "\n",
    "def process_video_and_csv(video_path, csv_path):\n",
    "    # Read the video\n",
    "    cap = cv2.VideoCapture(video_path)\n",
    "\n",
    "    # Create a list to store frames\n",
    "    frames = []\n",
    "\n",
    "    frame_count = 0\n",
    "    while cap.isOpened() and frame_count < 29:\n",
    "        ret, frame = cap.read()\n",
    "        if not ret:\n",
    "            break\n",
    "        frame = cv2.resize(frame, (64, 64))\n",
    "        frames.append(frame)\n",
    "        frame_count += 1\n",
    "\n",
    "    cap.release()\n",
    "\n",
    "    # Read the CSV file and skip the first two columns\n",
    "    csv_data = pd.read_csv(csv_path)\n",
    "    df = csv_data.iloc[:29, 2:]\n",
    "    # Return the frames and CSV data\n",
    "    return np.array(frames), np.array(df)\n",
    "\n",
    "# Example usage\n",
    "# video_path = 'D:/FYP/Data/video_frames/Data/1sec_videos/salman/video_002.mp4'\n",
    "# csv_path = 'D:/FYP/Data/video_frames/Data/CSV/salman/video_007.csv'\n",
    "# video_path = 'D:/FYP/Data/video_frames/junaid.mp4'\n",
    "# csv_path = 'D:/FYP/Data/video_frames/lstm_network/test/juanid/junaid.csv'\n",
    "\n",
    "conv3d_input, lstm_input = process_video_and_csv(video_path, csv_path)\n",
    "print(\"Number of frames:\",conv3d_input.shape)\n",
    "print(\"CSV data:\")\n",
    "print(lstm_input.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 38ms/step\n",
      "[[0.08007735 0.350006   0.5006316  0.06928503]]\n",
      "4\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Prediction code\n",
    "def predict_with_fused_model(fused_model, conv3d_input, lstm_input):\n",
    "    conv3d_input = np.expand_dims(conv3d_input, axis=0)\n",
    "    lstm_input = np.expand_dims(lstm_input, axis=0)\n",
    "    predictions = fused_model.predict([conv3d_input, lstm_input])\n",
    "    return predictions\n",
    "\n",
    "\n",
    "\n",
    "predictions = predict_with_fused_model(fused_model, conv3d_input, lstm_input)\n",
    "print(predictions)\n",
    "print(len(predictions[0]))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = predictions.squeeze().tolist()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The Probability of ali is 8.01%\n",
      "The Probability of huzaifa is 35.00%\n",
      "The Probability of junaid is 50.06%\n",
      "The Probability of salman is 6.93%\n"
     ]
    }
   ],
   "source": [
    "for i in range(len(predictions)):\n",
    "    print(f\"The Probability of {class_label[i]} is {predictions[i]*100:.2f}%\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tf",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.16"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}